Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
| Name | Class |
|---|---|
| Cornell University | OTHER |
| University of Waterloo | OTHER |
| University of Tasmania | OTHER |
Not provided
Not provided
Not provided
Not provided
This study will test whether cigarette package inserts (i.e., small printed leaflets inside cigarette packs) with messages about the benefits of cessation and tips for quitting can help smokers quit. To do this, the investigators will conduct a 2 X 2 between-subject experiment in which 380 smokers will be randomized into one of four labeling groups: 1. no inserts or pictorial health warning labels (HWLs); 2. inserts only; 3. pictorial HWLs only; 4. inserts & pictorial HWLs. Smokers will be given a 14-day supply of their preferred cigarette brand with packs labeled according to their experimental group. Participants will answer a brief survey at the end of each day and four other times each day, using ecological momentary assessment approaches. The investigators will study whether smokers in each group experience different psychological responses and behaviors associated with smoking cessation.
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
| Label | Type | Description | Intervention Names |
|---|---|---|---|
| Standard cigarette packs (control) | No Intervention | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with the only alteration being the small health warning message on the side of the pack, whose textual content will be the same as that used for the pictorial warning label conditions. | |
| Cigarette packs with inserts only | Experimental | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include inserts with four different rotating messages to promote response efficacy beliefs (2 inserts on the benefits of cessation) or self-efficacy to quit (2 inserts with cessation tips). |
|
| Cigarette packs with pictorial warnings only | Experimental | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include four different rotating pictorial warnings showing the consequences of smoking and that cover 50% of the front and back of the pack. |
|
| Cigarette packs with inserts and pictorial warnings | Experimental | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include inserts with four rotating efficacy messages (see description above) and four rotating pictorial warnings (see above). |
|
| Name | Type | Description | Arm Group Labels | Other Names |
|---|---|---|---|---|
| Inserts with efficacy messages | Other | Four rotating inserts (i.e., small, printed messages) with messages about cessation benefits (i.e., response efficacy) and tips to quit (i.e., self-efficacy messages) will be placed inside cigarette packs. |
| Measure | Description | Time Frame |
|---|---|---|
| Self-efficacy to Quit Smoking | A single question measured on a continuous scale (1="Not at all", worse outcome; 7="Extremely", better outcome) to assess strength of confidence to quit smoking, with average scores (range 1-7) estimated. | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
| Self-efficacy to Cut Down on Smoking | A single question measured on a continuous scale (1="Not at all", worse outcome; 7="Extremely", better outcome) to assess strength of confidence to cut down on the number of cigarettes smoked, with average scores (range=1-7) estimated. | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
| Worry About Harms From Smoking | A single question measured on a continuous scale to assess strength of worry about harms from smoking (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range=1-7) estimated. | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
| Strength of Feeling About Smoking | A single question measured on a continuous scale to assess strength of positive to negative feelings about smoking (1="very bad", better outcome; 7="very good", worse outcome), with average scores (range=1-7) estimated | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
| Extent of Motivation to Quit Smoking |
| Measure | Description | Time Frame |
|---|---|---|
| Strength of Hopefulness About Quitting | A single question measured on a continuous scale to assess strength of feeling hopeful about quitting (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range 1-7) estimated. | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
| Measure | Description | Time Frame |
|---|---|---|
| Conversation Partners | Nominal variable indicating the types of people with whom people talked about smoking and cessation in the prior 24 hours | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
Inclusion Criteria:
Exclusion Criteria:
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
| Name | Affiliation | Role |
|---|---|---|
| James F Thrasher, PhD | University of South Carolina | Principal Investigator |
| Facility | Status | City | State | ZIP | Country | Contacts |
|---|---|---|---|---|---|---|
| Cornell University | Ithaca | New York | 14853 | United States | ||
| University of South Carolina |
We are likely to make participant data to researchers who make specific requests and have specific data analysis plans that we approve.
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
| ID | Title | Description |
|---|---|---|
| FG000 | Standard Cigarette Packs (Control) | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with the only alteration being the small health warning message on the side of the pack, whose textual content will be the same as that used for the pictorial warning label conditions. |
| FG001 | Cigarette Packs With Inserts Only | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include inserts with four different rotating messages to promote response efficacy beliefs (2 inserts on the benefits of cessation) or self-efficacy to quit (2 inserts with cessation tips). Inserts with efficacy messages: Four rotating inserts (i.e., small, printed messages) with messages about cessation benefits (i.e., response efficacy) and tips to quit (i.e., self-efficacy messages) will be placed inside cigarette packs. |
| FG002 | Cigarette Packs With Pictorial Warnings Only | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include four different rotating pictorial warnings showing the consequences of smoking and that cover 50% of the front and back of the pack. Pictorial warnings on packs: Four rotating pictorial warnings illustrating the harms of smoking will be printed on labels that will placed on and cover approximately 50% of the cigarette packs. |
| FG003 | Cigarette Packs With Inserts and Pictorial Warnings | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include inserts with four rotating efficacy messages (see description above) and four rotating pictorial warnings (see above). Inserts with efficacy messages: Four rotating inserts (i.e., small, printed messages) with messages about cessation benefits (i.e., response efficacy) and tips to quit (i.e., self-efficacy messages) will be placed inside cigarette packs. Pictorial warnings on packs: Four rotating pictorial warnings illustrating the harms of smoking will be printed on labels that will placed on and cover approximately 50% of the cigarette packs. |
| Title | Milestones | Reasons Not Completed | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Overall Study |
|
|
The analysis population is those participants who were included in the analytic sample.
Not provided
| ID | Title | Description |
|---|---|---|
| BG000 | Standard Cigarette Packs (Control) | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with the only alteration being the small health warning message on the side of the pack, whose textual content will be the same as that used for the pictorial warning label conditions. |
| BG001 |
| Units | Counts |
|---|---|
| Participants |
|
| Title | Description | Population Description | Parameter Type | Dispersion Type | Unit of Measure | Calculate Percentage | Denominator Units Selected | Denominators | Classes |
|---|---|---|---|---|---|---|---|---|---|
| Age, Customized | Count of Participants |
| Type | Title | Description | Population Description | Reporting Status | Anticipated Posting Date | Parameter Type | Dispersion Type | Unit of Measure | Calculate Percentage | Time Frame | Units Analyzed | Denominator Units Selected | Arm/Group Information | Denominators | Classes | Analyses | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Primary | Self-efficacy to Quit Smoking | A single question measured on a continuous scale (1="Not at all", worse outcome; 7="Extremely", better outcome) to assess strength of confidence to quit smoking, with average scores (range 1-7) estimated. | Analyses include repeated observations from daily cigarette surveys. Analytic samples for H6 differ due to some missing data for each moderating variable: education (control=101; insert only=84; pictorial warnings only=87; insert + pictorial warnings=88); health literacy (control=101; insert only=86; pictorial warnings only=90; insert + pictorial warnings=88); and delayed discounting (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88). | Posted | Mean | Standard Deviation | score on a scale | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
|
Adverse events were assessed during the 14 day study participation period.
Adverse events were detected through the standardized study protocol of phoning participants when they stopped recording cigarettes and responding to evening surveys. Through this approach, we detected two adverse events that required hospitalization and may have been caused by long-term smoking.
Not provided
| ID | Title | Description | Deaths (Affected) | Deaths (At Risk) | Serious Events (Affected) | Serious Events (At Risk) | Other Events (Affected) | Other Events (At Risk) |
|---|---|---|---|---|---|---|---|---|
| EG000 | Standard Cigarette Packs (Control) | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with the only alteration being the small health warning message on the side of the pack, whose textual content will be the same as that used for the pictorial warning label conditions. |
| Term | Organ System | Source Vocabulary | Assessment Type | Notes | Statistical Information |
|---|---|---|---|---|---|
| Heart Attack | Cardiac disorders | Systematic Assessment | One participant in the inserts only condition was found to have been hospitalized after experiencing a heart attack when the study team followed up with him to determine why he had stopped logging cigarettes and surveys. |
Not provided
Not provided
| Title | Organization | Phone | Extension | |
|---|---|---|---|---|
| Dr. James F. Thrasher | University of South Carolina | 803.777.4862 | thrasher@sc.edu |
Not provided
| Type | Includes Protocol | Includes SAP | Includes ICF | Document Label | Document Date | Document Uploaded Date | Document File Name |
|---|---|---|---|---|---|---|---|
| Prot | Yes | No | No | Study Protocol | Jun 23, 2022 | Jun 24, 2022 | Prot_000.pdf |
| SAP | No | Yes | No | Statistical Analysis Plan | Dec 13, 2022 | Dec 13, 2022 | SAP_001.pdf |
| ICF | No | No | Yes | Informed Consent Form | Jul 21, 2020 | Jun 23, 2022 | ICF_002.pdf |
Not provided
| ID | Term |
|---|---|
| D016540 | Smoking Cessation |
| ID | Term |
|---|---|
| D015438 | Health Behavior |
| D001519 | Behavior |
Not provided
Not provided
In a 2 X 2 between subject design, smokers will be randomized into one of four cigarette package labeling conditions: 1. no inserts or pictorial HWLs; 2. inserts only; 3. pictorial HWLs only; 4. inserts & pictorial HWLs
Not provided
Not provided
Not provided
| Pictorial warnings on packs | Other | Four rotating pictorial warnings illustrating the harms of smoking will be printed on labels that will placed on and cover approximately 50% of the cigarette packs. |
|
A single question measured on a continuous scale to assess the extent of motivation to quit smoking (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range=1-7) estimated. |
| Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
| Talk About Smoking Cessation | For each day, a single binary variable was derived to indicate whether the participant talked about either smoking cessation or smoking-related harms in the prior 24 hours, with the average daily percentage of this outcome estimated. Experiencing the event of talking about cessation or harms is a better outcome, as it is a predictor of cessation attempts. | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
| Foregoing a Cigarette | For each day, responses to two questions were combined to create a binary variable for whether, in the prior 24 hours, the participant a) did not smoke a cigarette when they normally would, and/or b) stubbed out a cigarette before finishing it, with the average daily percentage of this outcome estimated. Experiencing the event of forgoing/stubbing out is a better outcome, as it is a predictor of cessation attempts. | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
| Extent of Cognitive Elaboration of Smoking Benefits | A single question measured on a continuous scale to assess the frequency of thinking about the positive things about smoking (1="Not at all", better outcome; 7="Extremely", worse outcome), with average scores (range 1-7) estimated. | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
| Extent of Cognitive Elaboration of Smoking Harms | A single question measured on a continuous scale to assess the frequency of thinking about the harms from smoking (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range 1-7) estimated. | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
| Extent of Cognitive Elaboration of Cessation Benefits | A single question measured on a continuous scale to assess the frequency of thinking about the potential benefits from cessation (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range 1-7) estimated. | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
| Strength of Beliefs About Benefits of Smoking Cessation (i.e., Response Efficacy) | A single question measured on a continuous scale to assess the strength of perceived benefits from cessation (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range 1-7) estimated. | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
| Strength of Perceived Susceptibility to Smoking Harms | A single question measured on a continuous scale to assess the perceived likelihood of suffering harms from smoking (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range 1-7) estimated. | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
| Expressed Reactance Against Health Messages | a. Binary measure for whether the participant reported talking about how health information on packs is either useless or not believable in the prior 24 hours, with the average daily percentage of this outcome estimated. Experiencing the event is a worse outcome, as it suggests defensive responding to labels. | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
| Satisfaction From Smoking | Single question measured on a continuous scale that assesses strength of satisfaction from smoking (1="Not at all", better outcome; 7="Extremely", worse outcome), with average scores (range 1-7) estimated. | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
| Cigarettes Logged | Continuous variable summing the total number of cigarettes logged over the study period. Lower values represent a better outcome (i.e., fewer cigarettes smoked). | Participants logged each cigarette smoked over the 14 day study period, with analyses limited to reports from days 2-14. Day 1 and 15 were dropped due to varying coverage of the day depending on study orientation time (Day 1) and ending (Day 15). |
| Conversation Topics | Nominal variable indicating the topics of conversation around smoking in the prior 24 hours | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
| Columbia |
| South Carolina |
| 29208 |
| United States |
| Withdrawal by Subject |
|
| Protocol Violation |
|
| Not literate enough to self-administer study surveys |
|
| Lost study phone the first day of study participation |
|
| Cigarette Packs With Inserts Only |
Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include inserts with four different rotating messages to promote response efficacy beliefs (2 inserts on the benefits of cessation) or self-efficacy to quit (2 inserts with cessation tips). Inserts with efficacy messages: Four rotating inserts (i.e., small, printed messages) with messages about cessation benefits (i.e., response efficacy) and tips to quit (i.e., self-efficacy messages) will be placed inside cigarette packs. |
| BG002 | Cigarette Packs With Pictorial Warnings Only | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include four different rotating pictorial warnings showing the consequences of smoking and that cover 50% of the front and back of the pack. Pictorial warnings on packs: Four rotating pictorial warnings illustrating the harms of smoking will be printed on labels that will placed on and cover approximately 50% of the cigarette packs. |
| BG003 | Cigarette Packs With Inserts and Pictorial Warnings | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include inserts with four rotating efficacy messages (see description above) and four rotating pictorial warnings (see above). Inserts with efficacy messages: Four rotating inserts (i.e., small, printed messages) with messages about cessation benefits (i.e., response efficacy) and tips to quit (i.e., self-efficacy messages) will be placed inside cigarette packs. Pictorial warnings on packs: Four rotating pictorial warnings illustrating the harms of smoking will be printed on labels that will placed on and cover approximately 50% of the cigarette packs. |
| BG004 | Total | Total of all reporting groups |
| Participants |
|
| Sex: Female, Male | Count of Participants | Participants |
|
| Race (NIH/OMB) | Count of Participants | Participants |
|
| Region of Enrollment | Number | participants |
|
| Educational Attainment | This measure is the dichotomized version of our original question and is what we used when assessing moderation of insert effects (H6a, H6b). | Number | participants |
|
| Health Literacy | Health Literacy was assessed with the 6-item Newest Vital Sign, which categorizes people as having adequate, possibly limited or limited health literacy. For our assessment of moderation of insert effects by health literacy (i.e., H6a, H6b), we created a dichotomous measure by combining the categories of possibly limited or limited. | Number | participants |
|
| Self-efficacy to quit smoking | This 3-item scale with Likert response options was averaged across the three items and has a range of 1-5, where higher values represent a better outcome (i.e., greater confidence to quit smoking). | Mean | Standard Deviation | units on a scale |
|
| Delayed Discounting | This measure assessed valuation of present rewards over future rewards, in which participants answer 5 questions with branching logic. Final values range from .0001 to 24 with a median of .0095, which we used to classify participants as having high vs. low delayed discounting. Low delayed discounting is indicative of a "better" outcome because people classified in this group are more likely to value future rewards (e.g., living longer) over present rewards (e.g., nicotine dosing), whereas the opposite is true for those with high delayed discounting. | Mean | Standard Deviation | units on a scale |
|
Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with the only alteration being the small health warning message on the side of the pack, whose textual content will be the same as that used for the pictorial warning label conditions. |
| OG001 | Cigarette Packs With Inserts Only | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include inserts with four different rotating messages to promote response efficacy beliefs (2 inserts on the benefits of cessation) or self-efficacy to quit (2 inserts with cessation tips). Inserts with efficacy messages: Four rotating inserts (i.e., small, printed messages) with messages about cessation benefits (i.e., response efficacy) and tips to quit (i.e., self-efficacy messages) will be placed inside cigarette packs. |
| OG002 | Cigarette Packs With Pictorial Warnings Only | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include four different rotating pictorial warnings showing the consequences of smoking and that cover 50% of the front and back of the pack. Pictorial warnings on packs: Four rotating pictorial warnings illustrating the harms of smoking will be printed on labels that will placed on and cover approximately 50% of the cigarette packs. |
| OG003 | Cigarette Packs With Inserts and Pictorial Warnings | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include inserts with four rotating efficacy messages (see description above) and four rotating pictorial warnings (see above). Inserts with efficacy messages: Four rotating inserts (i.e., small, printed messages) with messages about cessation benefits (i.e., response efficacy) and tips to quit (i.e., self-efficacy messages) will be placed inside cigarette packs. Pictorial warnings on packs: Four rotating pictorial warnings illustrating the harms of smoking will be printed on labels that will placed on and cover approximately 50% of the cigarette packs. |
|
|
|
| Primary | Self-efficacy to Cut Down on Smoking | A single question measured on a continuous scale (1="Not at all", worse outcome; 7="Extremely", better outcome) to assess strength of confidence to cut down on the number of cigarettes smoked, with average scores (range=1-7) estimated. | Analyses include repeated observations from daily cigarette surveys. Analytic samples for H6 differ due to some missing data for each moderating variable: education (control=101; insert only=84; pictorial warnings only=87; insert + pictorial warnings=88); health literacy (control=101; insert only=86; pictorial warnings only=90; insert + pictorial warnings=88); and delayed discounting (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88). | Posted | Mean | Standard Deviation | score on a scale | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
|
|
|
|
| Primary | Worry About Harms From Smoking | A single question measured on a continuous scale to assess strength of worry about harms from smoking (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range=1-7) estimated. | Analyses include repeated observations collected from daily cigarette surveys. The number of participants in the analytic sample is slightly smaller for H5 (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88) due to missing data from the baseline survey that provides information for the moderating variables. | Posted | Mean | Standard Deviation | score on a scale | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
|
|
|
|
| Primary | Strength of Feeling About Smoking | A single question measured on a continuous scale to assess strength of positive to negative feelings about smoking (1="very bad", better outcome; 7="very good", worse outcome), with average scores (range=1-7) estimated | Analyses include repeated observations collected from daily cigarette surveys. The number of participants in the analytic sample is slightly smaller for H5 (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88) due to missing data from the baseline survey that provides information for the moderating variables. | Posted | Mean | Standard Deviation | score on a scale | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
|
|
|
|
| Primary | Extent of Motivation to Quit Smoking | A single question measured on a continuous scale to assess the extent of motivation to quit smoking (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range=1-7) estimated. | Analyses include repeated observations from daily cigarette surveys. Due to missing data on moderators, the analytic sample differs for H5 (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88) and for each H6 moderating variable: education (n=101, 84, 87, and 88, respectively); health literacy (n=101, 86, 90, and 88, respectively); and delayed discounting (n=101, 84, 88, and 88, respectively). | Posted | Mean | Standard Deviation | score on a scale | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
|
|
|
|
| Primary | Talk About Smoking Cessation | For each day, a single binary variable was derived to indicate whether the participant talked about either smoking cessation or smoking-related harms in the prior 24 hours, with the average daily percentage of this outcome estimated. Experiencing the event of talking about cessation or harms is a better outcome, as it is a predictor of cessation attempts. | Analyses include repeated observations from evening reports. Due to missing data on moderators, the analytic sample differs for H5 (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88) and for each H6 moderating variable: education (n=101, 85, 87, and 88, respectively); health literacy (n=101, 87, 90, and 88, respectively); and delayed discounting (n=101, 85, 88, and 88, respectively). | Posted | Mean | Standard Deviation | percentage of days (events/days) | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
|
|
|
|
| Primary | Foregoing a Cigarette | For each day, responses to two questions were combined to create a binary variable for whether, in the prior 24 hours, the participant a) did not smoke a cigarette when they normally would, and/or b) stubbed out a cigarette before finishing it, with the average daily percentage of this outcome estimated. Experiencing the event of forgoing/stubbing out is a better outcome, as it is a predictor of cessation attempts. | Analyses include repeated observations from evening reports. Due to missing data on moderators, the analytic sample differs for H5 (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88) and for each H6 moderating variable: education (n=101, 85, 87, and 88, respectively); health literacy (n=101, 87, 90, and 88, respectively); and delayed discounting (n=101, 85, 88, and 88, respectively). | Posted | Mean | Standard Error | percentage of days (events/days) | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
|
|
|
|
| Secondary | Strength of Hopefulness About Quitting | A single question measured on a continuous scale to assess strength of feeling hopeful about quitting (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range 1-7) estimated. | Analyses include repeated observations from daily cigarette surveys. Analytic samples for H6 differ due to some missing data for each moderating variable: education (control=101; insert only=84; pictorial warnings only=87; insert + pictorial warnings=88); health literacy (control=101; insert only=86; pictorial warnings only=90; insert + pictorial warnings=88); and delayed discounting (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88). | Posted | Mean | Standard Deviation | score on a scale | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
|
|
|
|
| Secondary | Extent of Cognitive Elaboration of Smoking Benefits | A single question measured on a continuous scale to assess the frequency of thinking about the positive things about smoking (1="Not at all", better outcome; 7="Extremely", worse outcome), with average scores (range 1-7) estimated. | Analyses include repeated observations collected from evening surveys. | Posted | Mean | Standard Deviation | score on a scale | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
|
|
|
|
| Secondary | Extent of Cognitive Elaboration of Smoking Harms | A single question measured on a continuous scale to assess the frequency of thinking about the harms from smoking (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range 1-7) estimated. | Analyses include repeated observations collected from daily cigarette surveys or evening reports, depending on the outcome. The number of participants in the analytic sample is slightly smaller for H5 (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88) due to missing data from the baseline survey that provides information for the moderating variables. | Posted | Mean | Standard Deviation | score on a scale | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
|
|
|
|
| Secondary | Extent of Cognitive Elaboration of Cessation Benefits | A single question measured on a continuous scale to assess the frequency of thinking about the potential benefits from cessation (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range 1-7) estimated. | Analyses include repeated observations from evening reports. Analytic samples for H6 differ due to some missing data for each moderating variable: education (control=101; insert only=84; pictorial warnings only=87; insert + pictorial warnings=88); health literacy (control=101; insert only=86; pictorial warnings only=90; insert + pictorial warnings=88); and delayed discounting (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88). | Posted | Mean | Standard Deviation | score on a scale | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
|
|
|
|
| Secondary | Strength of Beliefs About Benefits of Smoking Cessation (i.e., Response Efficacy) | A single question measured on a continuous scale to assess the strength of perceived benefits from cessation (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range 1-7) estimated. | Analyses include repeated observations from evening reports. Analytic samples for H6 differ due to some missing data for each moderating variable: education (control=101; insert only=84; pictorial warnings only=87; insert + pictorial warnings=88); health literacy (control=101; insert only=86; pictorial warnings only=90; insert + pictorial warnings=88); and delayed discounting (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88). | Posted | Mean | Standard Deviation | Score on a scale | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
|
|
|
|
| Secondary | Strength of Perceived Susceptibility to Smoking Harms | A single question measured on a continuous scale to assess the perceived likelihood of suffering harms from smoking (1="Not at all", worse outcome; 7="Extremely", better outcome), with average scores (range 1-7) estimated. | Analyses include repeated observations collected from evening reports. The number of participants in the analytic sample is slightly smaller for H5 (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88) due to missing data from the baseline survey that provides information for the moderating variable. | Posted | Mean | Standard Deviation | score on a scale | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
|
|
|
|
| Secondary | Expressed Reactance Against Health Messages | a. Binary measure for whether the participant reported talking about how health information on packs is either useless or not believable in the prior 24 hours, with the average daily percentage of this outcome estimated. Experiencing the event is a worse outcome, as it suggests defensive responding to labels. | Analyses include repeated observations collected from daily cigarette surveys or evening reports, depending on the outcome. The number of participants in the analytic sample is slightly smaller for H5 (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88) due to missing data from the baseline survey that provides information for the moderating variables. | Posted | Mean | Standard Deviation | percentage of days (events/days) | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. |
|
|
|
|
| Secondary | Satisfaction From Smoking | Single question measured on a continuous scale that assesses strength of satisfaction from smoking (1="Not at all", better outcome; 7="Extremely", worse outcome), with average scores (range 1-7) estimated. | Analyses include repeated observations collected from daily cigarette surveys. | Posted | Mean | Standard Deviation | score on a scale | Evaluated approximately 4-5 times a day over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying coverage of the day depending on study orientation time. |
|
|
|
|
| Secondary | Cigarettes Logged | Continuous variable summing the total number of cigarettes logged over the study period. Lower values represent a better outcome (i.e., fewer cigarettes smoked). | Analyses assess the sum of all cigarettes participants logged from days 2-14 of the study. Due to missing data on moderators, the analytic sample differs for H5 (control=101; insert only=84; pictorial warnings only=88; insert + pictorial warnings=88) and for each H6 moderating variable: education (n=101, 85, 87, and 88, respectively); health literacy (n=101, 87, 90, and 88, respectively); and delayed discounting (n=101, 85, 88, and 88, respectively). | Posted | Mean | Standard Deviation | cigarettes | Participants logged each cigarette smoked over the 14 day study period, with analyses limited to reports from days 2-14. Day 1 and 15 were dropped due to varying coverage of the day depending on study orientation time (Day 1) and ending (Day 15). |
|
|
|
|
| Other Pre-specified | Conversation Partners | Nominal variable indicating the types of people with whom people talked about smoking and cessation in the prior 24 hours | Not Posted | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. | Participants |
| Other Pre-specified | Conversation Topics | Nominal variable indicating the topics of conversation around smoking in the prior 24 hours | Not Posted | Evaluated once each day (evening survey) over the 14 day study period, with analyses limited to days 2-14. Day 1 was dropped due to varying intervention exposure and coverage of the day, depending on study orientation time. | Participants |
| 0 |
| 108 |
| 0 |
| 108 |
| 0 |
| 108 |
| EG001 | Cigarette Packs With Inserts Only | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include inserts with four different rotating messages to promote response efficacy beliefs (2 inserts on the benefits of cessation) or self-efficacy to quit (2 inserts with cessation tips). Inserts with efficacy messages: Four rotating inserts (i.e., small, printed messages) with messages about cessation benefits (i.e., response efficacy) and tips to quit (i.e., self-efficacy messages) will be placed inside cigarette packs. | 0 | 94 | 1 | 94 | 0 | 94 |
| EG002 | Cigarette Packs With Pictorial Warnings Only | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include four different rotating pictorial warnings showing the consequences of smoking and that cover 50% of the front and back of the pack. Pictorial warnings on packs: Four rotating pictorial warnings illustrating the harms of smoking will be printed on labels that will placed on and cover approximately 50% of the cigarette packs. | 0 | 94 | 1 | 94 | 0 | 94 |
| EG003 | Cigarette Packs With Inserts and Pictorial Warnings | Participants will be provided with a 2 week supply of their preferred brand of cigarettes, with packs altered to include inserts with four rotating efficacy messages (see description above) and four rotating pictorial warnings (see above). Inserts with efficacy messages: Four rotating inserts (i.e., small, printed messages) with messages about cessation benefits (i.e., response efficacy) and tips to quit (i.e., self-efficacy messages) will be placed inside cigarette packs. Pictorial warnings on packs: Four rotating pictorial warnings illustrating the harms of smoking will be printed on labels that will placed on and cover approximately 50% of the cigarette packs. | 0 | 91 | 0 | 91 | 0 | 91 |
|
| Mini stroke | Vascular disorders | Systematic Assessment | One participant in the pictorial warning only group was found to have experienced what he termed a "mini stroke" when the study team followed up with him to determine why he had stopped logging cigarettes and surveys. |
|
Not provided
Not provided
|
|
| Male |
|
| Asian |
|
| Native Hawaiian or Other Pacific Islander |
|
| Black or African American |
|
| White |
|
| More than one race |
|
| Unknown or Not Reported |
|
|
|
|
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.08 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -1.25 | Standard Error of the Mean | 0.71 | 2-Sided | 95 | -2.63 | 0.14 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H4 assessment, comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.39 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.60 | Standard Error of the Mean | 0.70 | 2-Sided | 95 | -1.97 | 0.77 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.61 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 0.51 | Standard Error of the Mean | 1.00 | 2-Sided | 95 | -1.45 | 2.47 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.90 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -0.13 | Standard Error of the Mean | 1.07 | 2-Sided | 95 | -2.23 | 1.97 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.04 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 2.05 | Standard Error of the Mean | 0.98 | 2-Sided | 95 | 0.13 | 3.97 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.64 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 0.67 | Standard Error of the Mean | 1.43 | 2-Sided | 95 | -2.14 | 3.48 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.92 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 0.15 | Standard Error of the Mean | 1.56 | 2-Sided | 95 | -2.91 | 3.20 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.11 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -2.25 | Standard Error of the Mean | 1.41 | 2-Sided | 95 | -5.02 | 0.52 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial warnings (averaged over insert) vs no pictorial warnings (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4) | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.39 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.97 | Standard Error of the Mean | 1.14 | 2-Sided | 95 | -3.22 | 1.27 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.92 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.16 | Standard Error of the Mean | 1.63 | 2-Sided | 95 | -3.35 | 3.04 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial health warning labels (HWLs) (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.99 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 0.01 | Standard Error of the Mean | 0.94 | 2-Sided | 95 | -1.83 | 1.86 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.17 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 1.85 | Standard Error of the Mean | 1.34 | 2-Sided | 95 | -0.78 | 4.48 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.41 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.48 | Standard Error of the Mean | 0.58 | 2-Sided | 95 | -1.61 | 0.66 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.08 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -1.44 | Standard Error of the Mean | 0.84 | 2-Sided | 95 | -3.08 | 0.20 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.52 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.52 | Standard Error of the Mean | 0.82 | 2-Sided | 95 | -2.14 | 1.09 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.75 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 0.35 | Standard Error of the Mean | 1.11 | 2-Sided | 95 | -1.82 | 2.51 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.64 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.74 | Standard Error of the Mean | 1.57 | 2-Sided | 95 | -3.82 | 2.33 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.91 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 0.14 | Standard Error of the Mean | 1.18 | 2-Sided | 95 | -2.17 | 2.45 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.50 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -0.84 | Standard Error of the Mean | 1.25 | 2-Sided | 95 | -3.29 | 1.60 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.007 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 3.14 | Standard Error of the Mean | 1.16 | 2-Sided | 95 | 0.86 | 5.41 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.52 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 1.09 | Standard Error of the Mean | 1.69 | 2-Sided | 95 | -2.22 | 4.40 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.86 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -0.33 | Standard Error of the Mean | 1.82 | 2-Sided | 95 | -3.89 | 3.23 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.61 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -0.86 | Standard Error of the Mean | 1.68 | 2-Sided | 95 | -4.15 | 2.43 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.43 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.19 | Standard Error of the Mean | 0.27 | 2-Sided | 95 | 0.77 | 1.85 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.60 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 1.18 | Standard Error of the Mean | 0.37 | 2-Sided | 95 | 0.63 | 2.20 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.48 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.25 | Standard Error of the Mean | 0.40 | 2-Sided | 95 | 0.67 | 2.33 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.62 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.80 | Standard Error of the Mean | 0.36 | 2-Sided | 95 | 0.34 | 1.91 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.079 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 3.04 | Standard Error of the Mean | 1.93 | 2-Sided | 95 | 0.88 | 10.52 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.10 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 2.10 | Standard Error of the Mean | 0.95 | 2-Sided | 95 | 0.87 | 5.12 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.52 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.38 | Standard Error of the Mean | 0.68 | 2-Sided | 95 | 0.52 | 3.61 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.54 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.32 | Standard Error of the Mean | 0.59 | 2-Sided | 95 | 0.55 | 3.17 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.33 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.88 | Standard Error of the Mean | 1.21 | 2-Sided | 95 | 0.53 | 6.61 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.55 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.53 | Standard Error of the Mean | 1.08 | 2-Sided | 95 | 0.38 | 6.11 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.44 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 0.61 | Standard Error of the Mean | 0.39 | 2-Sided | 95 | 0.18 | 2.14 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the multiple imputation models for pre-specified H1 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received inserts (insert only AND insert + pictorial warnings) with the groups that did not (pictorial warning only AND control). We expected that the outcome would be higher in the insert groups than in the no insert groups. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.2510 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.29 | Standard Error of the Mean | 0.28 | 2-Sided | 95 | 0.84 | 1.98 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the multiple imputation model for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.2822 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.27 | Standard Error of the Mean | 0.28 | 2-Sided | 95 | 0.82 | 1.95 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for multiple imputation models for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.3558 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 1.34 | Standard Error of the Mean | 0.42 | 2-Sided | 95 | 0.72 | 2.47 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the multiple imputation model for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.3291 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.36 | Standard Error of the Mean | 0.43 | 2-Sided | 95 | 0.73 | 2.51 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.7818 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.89 | Standard Error of the Mean | 0.39 | 2-Sided | 95 | 0.38 | 2.09 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.0912 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 2.87 | Standard Error of the Mean | 1.79 | 2-Sided | 95 | 0.84 | 9.73 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.1361 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.94 | Standard Error of the Mean | 0.87 | 2-Sided | 95 | 0.81 | 4.66 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the multiple imputation model for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.7200 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.19 | Standard Error of the Mean | 0.58 | 2-Sided | 95 | 0.46 | 3.07 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.3737 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.48 | Standard Error of the Mean | 0.66 | 2-Sided | 95 | 0.62 | 3.53 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.1345 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 2.58 | Standard Error of the Mean | 1.63 | 2-Sided | 95 | 0.75 | 8.90 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.3318 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.96 | Standard Error of the Mean | 1.36 | 2-Sided | 95 | 0.50 | 7.67 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.7017 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 0.79 | Standard Error of the Mean | 0.49 | 2-Sided | 95 | 0.23 | 2.70 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.025 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.89 | Standard Error of the Mean | 0.54 | 2-Sided | 95 | 1.09 | 3.30 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.44 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.37 | Standard Error of the Mean | 0.56 | 2-Sided | 95 | 0.62 | 3.06 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.15 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.80 | Standard Error of the Mean | 0.73 | 2-Sided | 95 | 0.81 | 3.97 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.76 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.19 | Standard Error of the Mean | 0.67 | 2-Sided | 95 | 0.39 | 3.58 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.77 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.79 | Standard Error of the Mean | 0.63 | 2-Sided | 95 | 0.16 | 3.80 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.17 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 2.18 | Standard Error of the Mean | 1.23 | 2-Sided | 95 | 0.72 | 6.61 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.22 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.46 | Standard Error of the Mean | 0.29 | 2-Sided | 95 | 0.14 | 1.57 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.02 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 3.45 | Standard Error of the Mean | 1.94 | 2-Sided | 95 | 1.15 | 10.37 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.04 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 5.45 | Standard Error of the Mean | 4.40 | 2-Sided | 95 | 1.12 | 26.53 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.56 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.59 | Standard Error of the Mean | 0.53 | 2-Sided | 95 | 0.10 | 3.47 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.46 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.55 | Standard Error of the Mean | 0.44 | 2-Sided | 95 | 0.11 | 2.68 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the pre-specified H1 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received inserts (insert only AND insert + pictorial warnings) with the groups that did not (pictorial warning only AND control). We expected that the outcome would be higher in the insert groups than in the no insert groups. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.0002 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 2.58 | Standard Error of the Mean | 0.66 | 2-Sided | 95 | 1.56 | 4.27 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.0135 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.89 | Standard Error of the Mean | 0.48 | 2-Sided | 95 | 1.14 | 3.12 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.2619 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.51 | Standard Error of the Mean | 0.56 | 2-Sided | 95 | 0.73 | 3.12 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.0474 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 2.07 | Standard Error of the Mean | 0.76 | 2-Sided | 95 | 1.01 | 4.24 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.8094 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.13 | Standard Error of the Mean | 0.58 | 2-Sided | 95 | 0.42 | 3.07 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.7978 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.83 | Standard Error of the Mean | 0.60 | 2-Sided | 95 | 0.20 | 3.44 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.1016 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 2.29 | Standard Error of the Mean | 1.15 | 2-Sided | 95 | 0.85 | 6.15 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.1570 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.45 | Standard Error of the Mean | 0.25 | 2-Sided | 95 | 0.15 | 1.36 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.0297 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 3.02 | Standard Error of the Mean | 1.53 | 2-Sided | 95 | 1.11 | 8.18 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.0213 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 5.28 | Standard Error of the Mean | 3.81 | 2-Sided | 95 | 1.28 | 21.73 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.9648 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.96 | Standard Error of the Mean | 0.78 | 2-Sided | 95 | 0.20 | 4.74 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.6512 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.72 | Standard Error of the Mean | 0.53 | 2-Sided | 95 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.08 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -1.67 | Standard Error of the Mean | 0.96 | 2-Sided | 95 | -3.55 | 0.21 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.55 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -0.57 | Standard Error of the Mean | 0.95 | 2-Sided | 95 | -2.42 | 1.29 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.38 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | 1.20 | Standard Error of the Mean | 1.36 | 2-Sided | 95 | -1.47 | 3.86 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.71 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -0.54 | Standard Error of the Mean | 1.44 | 2-Sided | 95 | -3.35 | 2.28 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.02 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 3.21 | Standard Error of the Mean | 1.34 | 2-Sided | 95 | 0.59 | 5.83 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.98 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -0.05 | Standard Error of the Mean | 1.94 | 2-Sided | 95 | -3.85 | 3.76 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.31 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -2.13 | Standard Error of the Mean | 2.08 | 2-Sided | 95 | -6.2 | 1.95 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordinal regression models that accounted for repeated measures at both the day- and individual-levels. | 0.68 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.8 | Standard Error of the Mean | 1.93 | 2-Sided | 95 | -4.58 | 2.97 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.5767 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Mean Difference (Net) | -0.14 | Standard Error of the Mean | 0.25 | 2-Sided | 95 | -0.64 | 0.70 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.63 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.32 | Standard Error of the Mean | 0.67 | 2-Sided | 95 | -1.64 | 0.99 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.79 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.25 | Standard Error of the Mean | 0.95 | 2-Sided | 95 | -1.62 | 2.12 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.4491 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.23 | Standard Error of the Mean | 0.31 | 2-Sided | 95 | -0.84 | 0.37 | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.8713 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.10 | Standard Error of the Mean | 0.59 | 2-Sided | 95 | -1.26 | 1.06 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.99 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.00 | Standard Error of the Mean | 0.85 | 2-Sided | 95 | -1.66 | 1.66 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.08 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.87 | Standard Error of the Mean | 0.50 | 2-Sided | 95 | -1.85 | 0.11 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.65 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.23 | Standard Error of the Mean | 0.49 | 2-Sided | 95 | -0.74 | 1.19 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.30 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.73 | Standard Error of the Mean | 0.71 | 2-Sided | 95 | -0.66 | 2.11 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.78 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.21 | Standard Error of the Mean | 0.75 | 2-Sided | 95 | -1.68 | 1.26 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.03 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 1.48 | Standard Error of the Mean | 0.69 | 2-Sided | 95 | 0.12 | 2.83 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.95 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.06 | Standard Error of the Mean | 1.01 | 2-Sided | 95 | -1.92 | 2.04 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.48 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.78 | Standard Error of the Mean | 1.09 | 2-Sided | 95 | -2.91 | 1.36 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.37 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.90 | Standard Error of the Mean | 1.00 | 2-Sided | 95 | -2.86 | 1.07 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the pre-specified H1 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received inserts (insert only AND insert + pictorial warnings) with the groups that did not (pictorial warning only AND control). We expected that the outcome would be higher in the insert groups than in the no insert groups. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.0192 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.73 | Standard Error of the Mean | 0.31 | 2-Sided | 95 | 0.12 | 1.34 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.1860 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.59 | Standard Error of the Mean | 0.45 | 2-Sided | 95 | -1.48 | 0.29 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.3863 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.39 | Standard Error of the Mean | 0.44 | 2-Sided | 95 | -0.49 | 1.26 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.5710 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.36 | Standard Error of the Mean | 0.64 | 2-Sided | 95 | -0.89 | 1.61 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.2797 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.73 | Standard Error of the Mean | 0.68 | 2-Sided | 95 | -2.06 | 0.60 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.0312 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 1.33 | Standard Error of the Mean | 0.62 | 2-Sided | 95 | 0.12 | 2.55 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.5086 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.60 | Standard Error of the Mean | 0.91 | 2-Sided | 95 | -1.19 | 2.39 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.9474 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.06 | Standard Error of the Mean | 0.98 | 2-Sided | 95 | -1.86 | 1.99 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.5018 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.60 | Standard Error of the Mean | 0.90 | 2-Sided | 95 | -2.36 | 1.16 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.34 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.53 | Standard Error of the Mean | 0.56 | 2-Sided | 95 | -1.63 | 0.57 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.76 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.17 | Standard Error of the Mean | 0.56 | 2-Sided | 95 | -1.26 | 0.92 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.15 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 1.10 | Standard Error of the Mean | 0.78 | 2-Sided | 95 | -0.42 | 2.63 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 1.00 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.003 | Standard Error of the Mean | 0.85 | 2-Sided | 95 | -1.66 | 1.66 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.01 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 1.90 | Standard Error of the Mean | 0.78 | 2-Sided | 95 | 0.38 | 3.43 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.58 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.62 | Standard Error of the Mean | 1.11 | 2-Sided | 95 | -1.56 | 2.80 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.39 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -1.06 | Standard Error of the Mean | 1.23 | 2-Sided | 95 | -3.47 | 1.35 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.45 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.85 | Standard Error of the Mean | 1.12 | 2-Sided | 95 | -3.05 | 1.35 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the pre-specified H1 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received inserts (insert only AND insert + pictorial warnings) with the groups that did not (pictorial warning only AND control). We expected that the outcome would be higher in the insert groups than in the no insert groups. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.1147 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.54 | Standard Error of the Mean | 0.34 | 2-Sided | 95 | -0.13 | 1.21 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.3721 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.44 | Standard Error of the Mean | 0.49 | 2-Sided | 95 | -1.41 | 0.53 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.8827 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.07 | Standard Error of the Mean | 0.49 | 2-Sided | 95 | -0.89 | 1.03 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.4335 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.54 | Standard Error of the Mean | 0.68 | 2-Sided | 95 | -0.80 | 1.87 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.4063 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.62 | Standard Error of the Mean | 0.74 | 2-Sided | 95 | -2.07 | 0.84 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.0132 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 1.69 | Standard Error of the Mean | 0.68 | 2-Sided | 95 | 0.35 | 3.03 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.3949 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 0.83 | Standard Error of the Mean | 0.98 | 2-Sided | 95 | -1.08 | 2.75 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.7145 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.39 | Standard Error of the Mean | 1.08 | 2-Sided | 95 | -2.50 | 1.72 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.6081 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.51 | Standard Error of the Mean | 0.98 | 2-Sided | 95 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4) | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.82 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.20 | Standard Error of the Mean | 0.90 | 2-Sided | 95 | -1.98 | 1.57 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.20 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 1.67 | Standard Error of the Mean | 1.31 | 2-Sided | 95 | -0.90 | 4.24 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation analysis are for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.8800 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | -0.05 | Standard Error of the Mean | 0.36 | 2-Sided | 95 | -0.77 | 0.66 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4) | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.8864 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Median Difference (Net) | 0.10 | Standard Error of the Mean | 0.72 | 2-Sided | 95 | -1.30 | 1.50 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects ordered logistic regression models that accounted for repeated measures at the individual level. | 0.2748 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Mean Difference (Net) | 1.12 | Standard Error of the Mean | 1.02 | 2-Sided | 95 | -0.89 | 3.12 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.03 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 2.86 | Standard Error of the Mean | 1.34 | 2-Sided | 95 | 1.14 | 7.18 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be lower in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.18 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 0.53 | Standard Error of the Mean | 0.27 | 2-Sided | 95 | 0.15 | 1.44 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be lower for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.99 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.01 | Standard Error of the Mean | 1.30 | 2-Sided | 95 | 0.08 | 12.57 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.86 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 0.90 | Standard Error of the Mean | 0.52 | 2-Sided | 95 | 0.29 | 2.82 | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the pre-specified H1 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received inserts (insert only AND insert + pictorial warnings) with the groups that did not (pictorial warning only AND control). We expected that the outcome would be lower in the insert groups than in the no insert groups. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.8045 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 1.10 | Standard Error of the Mean | 0.43 | 2-Sided | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.0205 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 2.53 | Standard Error of the Mean | 1.01 | 2-Sided | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results from the multiple imputation model are for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be lower in the insert and pictorial warning group than in the pictorial warning only group. | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.1918 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Odds Ratio (OR) | 0.53 | Standard Error of the Mean | 0.26 | 2-Sided | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be lower for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.99 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.03 | Standard Error of the Mean | 1.12 | 2-Sided | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are from the multiple imputation model for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Mixed Models Analysis | We estimated mixed-effects logistic regression models that accounted for repeated measures at the individual level. | 0.6873 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083. | Odds Ratio (OR) | 1.56 | Standard Error of the Mean | 1.54 | 2-Sided | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. |
| Other |
We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H2 assessment (see Statistical Analysis Plan, Model 1), comparing groups that received pictorial warnings (pictorial warning only AND insert + pictorial warnings) with the groups that did not (insert only AND control). We expected that the outcome would be higher in the pictorial warning groups than in the no pictorial warning groups. | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.95 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 1 | Standard Error of the Mean | 0.05 | 2-Sided | 95 | 0.90 | 1.11 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H3 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received inserts only. We expected that this outcome would be higher in the insert and pictorial warning group than in the insert only group. | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.85 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 0.99 | Standard Error of the Mean | 0.08 | 2-Sided | 95 | 0.85 | 1.14 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the pre-specified H4 assessment (see Statistical Analysis Plan, Model 1), comparing the group that received inserts and pictorial warnings with the group that received pictorial warnings only. We expected that this outcome would be higher in the insert and pictorial warning group than in the pictorial warning only group. | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.75 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 0.98 | Standard Error of the Mean | 0.07 | 2-Sided | 95 | 0.84 | 1.13 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5a assessment, which evaluates moderation of the pre-specified H2 assessment. We expected that the effect of pictorial HWLs (averaged over insert) vs no pictorial HWLs (averaged over insert) on the outcome would be stronger for higher vs lower self-efficacy (see Statistical Analysis Plan, Model 4) | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.78 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 0.97 | Standard Error of the Mean | 0.11 | 2-Sided | 95 | 0.78 | 1.20 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H5b assessment, which evaluates whether moderation of pictorial warnings effects would vary by whether the pictorial warnings included inserts or not. We expected that moderation effects of baseline self-efficacy would be larger for pictorial warnings only (vs control) than for pictorial warnings AND inserts (vs control) (see Statistical Analysis Plan, Model 4). | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.11 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 1.31 | Standard Error of the Mean | 0.22 | 2-Sided | 95 | 0.94 | 1.82 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower education (see Statistical Analysis Plan, Model 5). | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.93 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 0.99 | Standard Error of the Mean | 0.11 | 2-Sided | 95 | 0.80 | 1.22 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with higher than lower literacy (see Statistical Analysis Plan, Model 5). | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.86 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 0.98 | Standard Error of the Mean | 0.11 | 2-Sided | 95 | 0.78 | 1.23 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6a assessment, which evaluates moderation of the pre-specified H1 assessment. We expected that the effect of inserts (averaged over pictorial warnings) compared to no inserts (averaged over pictorial warnings) on the outcome would be stronger for those with lower than higher delayed discounting (see Statistical Analysis Plan, Model 5). | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.94 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 0.99 | Standard Error of the Mean | 0.11 | 2-Sided | 95 | 0.81 | 1.22 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline education would be stronger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.13 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 1.29 | Standard Error of the Mean | 0.21 | 2-Sided | 95 | 0.93 | 1.78 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline literacy would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.96 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 0.99 | Standard Error of the Mean | 0.18 | 2-Sided | 95 | 0.70 | 1.41 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |
| These results are for the intended (but not correctly pre-specified) H6b assessment, which evaluated whether the moderation of insert effects would vary by whether the inserts included pictorial warnings or not. We expected that moderation by baseline delay discounting would be larger for insert only (vs control) than for inserts AND pictorial warnings (vs control) (see Statistical Analysis Plan, Model 5). | Negative binomial regression | We estimated negative binomial regression models, as there is a single measure for each person and so there are no repeated measures. | 0.95 | The reported p-value is not adjusted. The a priori threshold for statistical significance uses a Bonferroni-adjusted critical value of p<0.05/6=0.0083 | Rate ratio | 0.99 | Standard Error of the Mean | 0.16 | 2-Sided | 95 | 0.72 | 1.36 | To evaluate the study hypothesis, a Wald test was used for the associated regression parameter. | Other | We conducted a two-sided test for mean (group 1) = mean (group 2) versus means not being equal. |