3 text messages per week and one interactive text per month.

Referral to a non-profit that connects families with community resources and provides families with case management.

The DialogPR is a caregiver report measure of shared reading quality. Score range from 0 to 30 with higher scores indicating more interactive reading.

This questionnaire was modified from the National Survey of Early Childhood Health and asks two questions about verbal and physical discipline. Scores range from 0 to 6 and higher scores indicate harsher discipline strategies.

The Physician-Parent Communication Survey is a caregiver reported measure of the caregiver-clinician relationship. We analyzed the interest subscale. The interest subscale includes 6 items are scored from 1 to 8. Scores from each item are averaged; thus total possible scores range from 1-8, with higher scores indicating stronger relationships.

Caregiver report of social needs and stressors was assessed via the Local Inventory of Needs and Knowledge (LINK) survey. Caregivers rate statements on social needs and stress from 0 to 4 with higher scores indicating greater needs and stress. Possible scores range from 0 to 16 with higher scores indicating greater needs and economic stress.

The Parental Stress Scale is an eighteen-item measure that assesses parental stress. Scores range from 18 to 90 with higher scores indicating greater stress.

The ScreenQ is a caregiver report measure of media use in children. Scores range from 0 to 26 with higher score indicating more media use.

In addition to ROR and text messages, participants will be referred to a county-based single point of entry system for referrals to community resources. This system simplifies access to poverty-reducing resources by creating a centralized access point, maintaining an updated data base of resources with existing capacity to support families, and providing case management.

Usual care including ROR: Usual care which includes ROR.

Text messages: 3 text messages per week and one interactive text per month.

Connection to community resources: Referral to a non-profit that connects families with community resources and provides families with case management.

Total of all reporting groups

Age of the child in the caregiver/child dyad.

Sex of the caregiver in the caregiver/child dyad.

Sex of the child in the caregiver/child dyad.

All participants self-identified as Latino/a/x as this was part of the inclusion criteria.

Health literacy score was assessed using the Short Assessment of Health Literacy (SAHL), and was administered in English or Spanish, based on the participant's language preference. The SAHL has 18 questions, with each question including (1) correctly reading a health-related word aloud, and (2) hearing 2 additional words, and choosing the one most closely related to the original health-related word.

Total possible scores range from 0-18, with scores 14 and below indicating low health literacy.

Child's social-emotional skills were assessed with the Devereux Early Childhood Assessment for Infants and Toddlers (DECA-I/T), which has 2 caregiver-report forms: an infant form (1-18 months) that includes Attachment/Relationships and Initiative subscales, and a toddler form (18-36 months) that includes Attachment/Relationships, Initiative, and Self-Regulation subscales. Scores are standardized on a T-distribution (M=50; SD=10). The Total Protective Factors score is calculated by adding subscale T-scores. T-scores of 60+ indicate an area strength, and T-scores of 40- indicate an area need.

Reading activities were assessed with the STIMQ2 reading scale, a parent-report measure. Subscales include Quantity, Diversity, and Quality. Total possible scores range from 0 to 19, with higher scores equating to higher quantity, diversity, and quality of shared reading.

Community resource participation was measures via the Survey of Income and Program Participation (SIPP), a self-reported measure of participation in different types of community resources (e.g., food/nutrition support, income support, childcare assistance, etc.). Possible scores range from 0 to 17, with higher scores indicating more involvement in community resources.

Caregiver-clinician relationship was assessed with the Modified Physician-Parent Communication Survey, Interest subscale. The interest subscale includes 6 items scored from 1 to 8. Item scores are then averaged. Total possible scores range from 1-8, with higher scores indicating stronger relationships.

Economic stress was assessed using the Local Inventory of Needs and Knowledge (LINK). Questions are answered on a 5-point Likert scale of "strongly disagree" (0) to "strongly agree" (4). Possible scores range from 0 to 16 with higher scores indicating greater needs and economic stress.

Caregiver stress was measured via the Parental Stress Scale (PSS), an 18-item scale. For each item, parents rate how much they agree or disagree using a 5-point Likert scale ranging from "strongly disagree" (1) to "strongly agree" (5). Total possible scores range from 18 to 90 with higher scores indicating more parental stress.

Child media use was measured using the Modified ScreenQ-I/T 2.0 (Infant/Toddler) version. There are 15 items, with total possible scores ranging from 0 to 26. Higher scores indicate more child media use.

In addition to ROR, participants will receive three text messages per week, plus one interactive follow-up message per month, for the study period with scheduled breaks.

Usual care including ROR: Usual care which includes ROR.

Text messages: 3 text messages per week and one interactive text per month.

In addition to ROR and text messages, participants will be referred to a county-based single point of entry system for referrals to community resources. This system simplifies access to poverty-reducing resources by creating a centralized access point, maintaining an updated data base of resources with existing capacity to support families, and providing case management.

Usual care including ROR: Usual care which includes ROR.

Text messages: 3 text messages per week and one interactive text per month.

Connection to community resources: Referral to a non-profit that connects families with community resources and provides families with case management.

Caregiver report of non-verbal communication and early expressive language assessed using the MacArthur Bates Communicative Development Inventories. The Words & Gestures short form is an 89-word vocabulary checklist with separate columns for comprehension and production for children 8-18 months. The Words & Sentences versions contain a 100-word productive vocabulary checklist and a question about combining words for children 16-30 months. Scores are summed and higher scores denote stronger skills.

Analyses for the 9-month and 18-month follow-up visits include participants who had baseline data and completed the MacArthur Bates Communicative Development Inventories at the 9-month and/or 18-month follow-up visit(s), respectively. Participants who did not participate in the 9-month follow-up were still welcome to participate in the 18-month follow-up.

Child's social-emotional skills were assessed with the Devereux Early Childhood Assessment for Infants and Toddlers (DECA-I/T), which has 2 caregiver-report forms: an infant form (1-18 months) that includes Attachment/Relationships and Initiative subscales, and a toddler form (18-36 months) that includes Attachment/Relationships, Initiative, and Self-Regulation subscales. Scores are standardized on a T-distribution (M=50; SD=10). The Total Protective Factors score is calculated by adding subscale T-scores. T-scores of 60+ indicate an area strength, and T-scores of 40- indicate an area need.

Analyses for the 9-month and 18-month follow-up visits include participants who had baseline data and completed the DECA at the 9-month and/or 18-month follow-up visit(s), respectively. Participants who did not participate in the 9-month follow-up were still welcome to participate in the 18-month follow-up.

The StimQ2 is a caregiver-reported measure of cognitive stimulation for children that includes domains on the home literacy environment (Reading scale) and responsiveness (Parental Verbal Responsiveness scale). The Reading scale includes 3 subdimensions (Book Reading Quantity, Diversity of Content, Book Reading Quality). Scores on the Reading scale range from 0 to 19. The Parental Verbal Responsiveness scale is composed of two subdimensions (Everyday Routines and Play and Pretend). Scores range from 0 to 15. Higher scores indicate more cognitive stimulation. The Reading scale was completed at baseline, the 9-month follow-up, and the 18-month follow-up. The Parent Verbal Responsiveness scale was completed at the 9-month follow-up and the18-month follow-up.

Analyses for the 9-month and 18-month follow-up visits include participants who had baseline data and completed the StimQ2 at the 9-month and/or 18-month follow-up visit(s), respectively. Participants who did not participate in the 9-month follow-up were still welcome to participate in the 18-month follow-up.

The DialogPR is a caregiver report measure of shared reading quality. Score range from 0 to 30 with higher scores indicating more interactive reading.

Analyses for the 9-month visit include participants who had baseline data and completed the DialogPR at the 9-month visit.

This questionnaire was modified from the National Survey of Early Childhood Health and asks two questions about verbal and physical discipline. Scores range from 0 to 6 and higher scores indicate harsher discipline strategies.

Analyses include participants who had baseline data and completed the caregiver discipline strategy measure at the 18-month follow-up visit.

The Parent Reading Belief Inventory is a caregiver reported measure of attitudes about reading with children that included positive affect (scores range from 0 to 33). Higher scores indicate more favorable attitudes and greater knowledge.

Analyses for the 9-month and 18-month follow-up visits include participants who had baseline data and completed the Parent Reading Belief Inventory (positive affect scale) at the 9-month and/or 18-month follow-up visit(s), respectively. Participants who did not participate in the 9-month follow-up were still welcome to participate in the 18-month follow-up.

Community resource participation was measured via the Survey of Income and Program Participation (SIPP), a series of questions that asks about caregiver and child participation in different types of community resources (e.g. food/nutrition support, income support, childcare assistance, etc.). Possible scores range from 0 to 17, with higher scores indicating more involvement in community resources.

Analyses for the 9-month and 18-month follow-up visits include participants who had baseline data and completed the SIPP at the 9-month and/or 18-month follow-up visit(s), respectively. Participants who did not participate in the 9-month follow-up were still welcome to participate in the 18-month follow-up.

The Physician-Parent Communication Survey is a caregiver reported measure of the caregiver-clinician relationship. We analyzed the interest subscale. The interest subscale includes 6 items are scored from 1 to 8. Scores from each item are averaged; thus total possible scores range from 1-8, with higher scores indicating stronger relationships.

Analyses for the 9-month and 18-month follow-up visits include participants who had baseline data and completed the Physician-Parent Communication Survey at the 9-month and/or 18-month follow-up visit(s), respectively. Participants who did not participate in the 9-month follow-up were still welcome to participate in the 18-month follow-up.

Caregiver report of social needs and stressors was assessed via the Local Inventory of Needs and Knowledge (LINK) survey. Caregivers rate statements on social needs and stress from 0 to 4 with higher scores indicating greater needs and stress. Possible scores range from 0 to 16 with higher scores indicating greater needs and economic stress.

Analyses for the 9-month and 18-month follow-up visits include participants who had baseline data and completed the LINK at the 9-month and/or 18-month follow-up visit(s), respectively. Participants who did not participate in the 9-month follow-up were still welcome to participate in the 18-month follow-up.

The Parental Stress Scale is an eighteen-item measure that assesses parental stress. Scores range from 18 to 90 with higher scores indicating greater stress.

Analyses for the 9-month and 18-month follow-up visits include participants who had baseline data and completed the Parental Stress Scale at the 9-month and/or 18-month follow-up visit(s), respectively. Participants who did not participate in the 9-month follow-up were still welcome to participate in the 18-month follow-up.

The ScreenQ is a caregiver report measure of media use in children. Scores range from 0 to 26 with higher score indicating more media use.

Analyses for the 9-month and 18-month follow-up visits include participants who had baseline data and completed the ScreenQ at the 9-month and/or 18-month follow-up visit(s), respectively. Participants who did not participate in the 9-month follow-up were still welcome to participate in the 18-month follow-up.

Caregiver's report of receipt of literacy promotion components during their previous healthcare visit.

Analyses for the 9-month and 18-month follow-up visits include participants who had baseline data and completed the 9-month and/or 18-month follow-up visit(s), respectively. Participants who did not participate in the 9-month follow-up were still welcome to participate in the 18-month follow-up.

Observation of parent-child interactions during shared reading and free play; observations will be coded based on DialogPR items

In addition to ROR, participants will receive three text messages per week, plus one interactive follow-up message per month, for the study period with scheduled breaks.

Usual care including ROR: Usual care which includes ROR.

Text messages: 3 text messages per week and one interactive text per month.

In addition to ROR and text messages, participants will be referred to a county-based single point of entry system for referrals to community resources. This system simplifies access to poverty-reducing resources by creating a centralized access point, maintaining an updated data base of resources with existing capacity to support families, and providing case management.

Usual care including ROR: Usual care which includes ROR.

Text messages: 3 text messages per week and one interactive text per month.

Connection to community resources: Referral to a non-profit that connects families with community resources and provides families with case management.

Analysis for 9-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple, potentially highly correlated primary outcomes, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment across primary outcomes and timepoints within domain (total=3 outcomes within the language development). Right skewed variables with minimum values of zero were log-transformed after adding one. minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for correlations due to repeated-measure effects for both active treatments versus control. The adjusted value for child communicative development was the value of the outcome minus the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and treatment fixed effects.

Analysis for 18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 2.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple, potentially highly correlated primary outcomes, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment across primary outcomes and timepoints within domain (total=3 outcomes within the language development). Right skewed variables with minimum values of zero were log-transformed after adding one. minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for correlations due to repeated-measure effects for both active treatments versus control. The adjusted value for child communicative development was the value of the outcome minus the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and treatment fixed effects.

Analysis for 18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple, potentially highly correlated primary outcomes, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment across primary outcomes and timepoints within domain (total=3 outcomes within the language development). Right skewed variables with minimum values of zero were log-transformed after adding one. minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for correlations due to repeated-measure effects for both active treatments versus control. The adjusted value for child communicative development was the value of the outcome minus the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and treatment fixed effects.

Analysis for 9-month follow-up, A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the social emotional domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=2 outcomes within the social emotional). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects for both active treatments versus control. The adjusted value for social emotional development was the value of the outcome minus the effect of baseline and the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and both treatment and baseline fixed effects.

Analysis for 18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 2.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the social emotional domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=2 outcomes within the social emotional). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects for both active treatments versus control. The adjusted value for social emotional development was the value of the outcome minus the effect of baseline and the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and both treatment and baseline fixed effects.

Analysis for 18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the social emotional domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=2 outcomes within the social emotional). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects for both active treatments versus control. The adjusted value for social emotional development was the value of the outcome minus the effect of baseline and the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and both treatment and baseline fixed effects.

Analysis for the Reading Scale at the 9-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the parental behavior domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=6 outcomes within the parental behavior). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects (as appropriate) for both active treatments versus control. The adjusted value for reading activities was the value of the outcome minus the effect of baseline and the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and both treatment and baseline fixed effects.

Analysis for the Reading Scale at the18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 2.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the parental behavior domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=6 outcomes within the parental behavior). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects (as appropriate) for both active treatments versus control. The adjusted value for reading activities was the value of the outcome minus the effect of baseline and the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and both treatment and baseline fixed effects.

Analysis for the Reading Scale at the 18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the parental behavior domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=6 outcomes within the parental behavior). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects (as appropriate) for both active treatments versus control. The adjusted value for reading activities was the value of the outcome minus the effect of baseline and the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and both treatment and baseline fixed effects.

Analysis for the Parent Verbal Responsiveness Scale at the 9-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 2.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the parental behavior domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=6 outcomes within the parental behavior). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects (as appropriate) for both active treatments versus control. The adjusted value for verbal responsiveness was the value of the outcome minus the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and treatment fixed effects.

Analysis for the Parent Verbal Responsiveness Scale at the 9-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the parental behavior domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=6 outcomes within the parental behavior). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects (as appropriate) for both active treatments versus control. The adjusted value for verbal responsiveness was the value of the outcome minus the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and treatment fixed effects.

Analysis for the Parent Verbal Responsiveness Scale at the 18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 2.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the parental behavior domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=6 outcomes within the parental behavior). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects (as appropriate) for both active treatments versus control. The adjusted value for verbal responsiveness was the value of the outcome minus the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and treatment fixed effects.

Analysis for the Parent Verbal Responsiveness Scale at the 18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the parental behavior domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=6 outcomes within the parental behavior). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects (as appropriate) for both active treatments versus control. The adjusted value for verbal responsiveness was the value of the outcome minus the best linear unbiased predictor (BLUP) estimated from a mixed linear model with random individual-specific intercepts and treatment fixed effects.

A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the parental behavior domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=6 outcomes within the parental behavior). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects (as appropriate) for both active treatments versus control. Dialogic reading behavior was measured at a single timepoint (9 months), so no adjustment was necessary.

A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

p-value is adjusted using the minP approach for potentially highly correlated outcomes. Then, p < 0.05 was used as an a priori threshold for statistical significance.

We followed the intent to treat principle, with participants included based on their group assignment. Because of the multiple correlated timepoints within the parental behavior domain, we used the step-down minP method (a re-sampling approach), to assess the effects of treatment (total=6 outcomes within the parental behavior). minP-adjusted p-values were calculated with ANOVA models comparing outcomes adjusted for baseline- and correlations due to repeated-measure effects (as appropriate) for both active treatments versus control. Discipline strategies were measured at a single timepoint (18 months), so no adjustment was necessary.

Analysis at the 18-month follow-up.

\A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 2.

The a priori threshold for statistical significance was p < .05.

Analysis at the 18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

The a priori threshold for statistical significance was p < .05.

Analysis at the18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 2.

The a priori threshold for statistical significance was p < .05.

Analysis at the 18-month follow-up.

\A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

The a priori threshold for statistical significance was p < .05.

Analysis at the 18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 2.

The a priori threshold for statistical significance was p < .05.

Analysis at the 18-month follow-up. A total of 630 parent-infant dyads (210 per arm) were needed to achieve 80% power to detect a 0.3 standard deviation (SD) difference assuming approximately 20% attrition. The null hypothesis was that there would be no differences between Arm 1 and Arm 3.

The a priori threshold for statistical significance was p < .05.