Stroke recognition was scored on a 0 - 9 point scale where 0 represents no correct answers regarding 9 scenarios and 9 represents perfect stroke recognition.

Perception of social norms is measured by the odds ratio of the responses to questions of participant agreement with others' influence to calling 911 if he/she were to see a stroke. Odds ratios measure the odds of responses, so higher odds ratios suggest greater odds of the positive change in social norms in the post-test compared to the pre-test. Questions: 1) Most people would call 911 if they were to see a stroke. 2) My family would want me to call 911 if I were to see a stroke. Given that participants within each church are more alike than participants between churches and the multiple time points, hierarchical models were used. Specifically, multilevel mixed-effects ordered logistic regression models with a fixed church-level intercept and a random participant level intercept were used to explore change between baseline and immediate post-test and baseline and delayed post-test social norms after accounting for the participants' church.

Perception of self-efficacy is measured by the odds ratios of the responses to questions of participant confidence in being able to identify and respond appropriately to a stroke. Odds ratios measure the odds of responses, so higher odds ratios suggest greater odds of positive self-efficacy change in the post-test compared to the pretest. Questions asking about self-efficacy were:1) I would be able to tell if someone is having a stroke and 2) I know what to do if I saw someone having a stroke. Given that participants within each church are more alike than participants between churches and multiple time points hierarchical models were used. Specifically, multilevel mixed-effects ordered logistic regression models with a fixed church-level intercept and a random participant level intercept were used to explore change between baseline and immediate post-test and baseline and delayed posttest self-efficacy after accounting for the participants' church.

Stroke attitude is measured by the odds ratio of participant's positive perception of calling 911 for stroke. Odds ratios measure the odds of responses, so higher odds ratios suggest greater odds of stroke attitude change in the post-test compared to pre-test. Stroke attitude questioners were: Q1) If I were to see signs of a stroke, calling 911 would be... (range "extremely pleasant" to "very unpleasant); and Q2) If a person has signs of a stroke, calling 911 right away could be... (range "very helpful" to "very harmful). Given that participants within each church are more alike than participants between churches and multiple time points, hierarchical models were used. Specifically, multilevel mixed-effects ordered logistic regression models with a fixed church-level intercept and a random participant level intercept were used to explore change between baseline and immediate post-test and baseline and delayed post-test stroke attitude after accounting for the participants' church.

The pre-test is conducted one week prior to the post-test. A higher score indicates greater behavioral intent. Behavioral intent is measured on a scale of 0 - 8, where 0 indicates no correct answers in responses to scenarios, and 8 indicates appropriate responses (calling 911 every time it is appropriate) to the scenarios presented.

Stroke recognition was scored on a 0 - 9 point scale where 0 represents no correct answers regarding 9 scenarios and 9 represents perfect stroke recognition.

Perception of social norms is measured by the odds ratio of the responses to questions of participant agreement with others' influence to calling 911 if he/she were to see a stroke. Odds ratios measure the odds of responses, so higher odds ratios suggest greater odds of the positive change in social norms in the post-test compared to the pre-test. Questions: 1) Most people would call 911 if they were to see a stroke. 2) My family would want me to call 911 if I were to see a stroke. Given that participants within each church are more alike than participants between churches and the multiple time points, hierarchical models were used. Specifically, multilevel mixed-effects ordered logistic regression models with a fixed church-level intercept and a random participant level intercept were used to explore change between baseline and immediate post-test and baseline and delayed post-test social norms after accounting for the participants' church.

Perception of self-efficacy is measured by the odds ratios of the responses to questions of participant confidence in being able to identify and respond appropriately to a stroke. Odds ratios measure the odds of responses, so higher odds ratios suggest greater odds of positive self-efficacy change in the post-test compared to the pretest. Questions asking about self-efficacy were:1) I would be able to tell if someone is having a stroke and 2) I know what to do if I saw someone having a stroke. Given that participants within each church are more alike than participants between churches and multiple time points hierarchical models were used. Specifically, multilevel mixed-effects ordered logistic regression models with a fixed church-level intercept and a random participant level intercept were used to explore change between baseline and immediate post-test and baseline and delayed posttest self-efficacy after accounting for the participants' church.

Stroke attitude is measured by the odds ratio of participant's positive perception of calling 911 for stroke. Odds ratios measure the odds of responses, so higher odds ratios suggest greater odds of stroke attitude change in the post-test compared to pre-test. Stroke attitude questioners were: Q1) If I were to see signs of a stroke, calling 911 would be... (range "extremely pleasant" to "very unpleasant); and Q2) If a person has signs of a stroke, calling 911 right away could be... (range "very helpful" to "very harmful). Given that participants within each church are more alike than participants between churches and multiple time points, hierarchical models were used. Specifically, multilevel mixed-effects ordered logistic regression models with a fixed church-level intercept and a random participant level intercept were used to explore change between baseline and immediate post-test and baseline and delayed post-test stroke attitude after accounting for the participants' church.

Program satisfaction is measured by percentage of participants that completed the program who answered on the post test: very satisfied or extremely satisfied on a questionnaire about the program.