Worldviews and Behavior: An Experimental Study

2024-10-30

2025-04-30

Our primary measure of consumption behavior is based on a menu choice that participants make at three points during the study: prior to watching the video during the in-person session, immediately after watching the video during the in-person session, and during the online survey that is sent to each participant seven days after the in-person session. These choices are shown in Sections 1B, 1I, and 2B of the attached survey. In each of these choices, participants select one of four available entrées and one of five protein options to go with their selected entrée. These choices are incentivized in that participants are entered into a lottery with a roughly 50% chance of winning a voucher to receive one of their three menu choices (i.e., their selected entrée and protein) at a local café. For our analysis, we will distill these choices into a binary measure of whether the participant selected a meat option (chicken, beef, bacon) or a non-meat option (tofu, none) in their choice of protein. In turn, we will measure both the immediate and longer-term effects of our treatment video on consumption behavior. The immediate effect will be measured using the difference between the responses to questions in sections 1I and 1B while the longer-term effect will be measured using the difference between the responses to questions 2B and 1B.

Our primary measure of worldview change is based on the degree of worldview change indicated in questions 1G and 2C of the attached survey. For our analysis, we will express participants’ responses on a 0 to 5 scale (with 0 representing “no change” and 5 representing “very large change”). The immediate effect of our treatment video will be measured using responses to the worldview question in Section 1G of our attached survey while the longer-term effect will be measured using responses to the same question in Section 2C. Our analysis of worldview change will also examine a potential relationship between worldview change and mindset flexibility. Here, “mindset flexibility” will be operationalized as the average score on the two mindset flexibility questions in section 1C, with responses to each question expressed on a 1 to 5 scale (with 1 representing “strongly disagree” and 5 representing “strongly agree”).

As a secondary measure of consumption, we will look at participants’ total self-reported meat consumption, including the number of times the participant ate “pork products” in the last week and the number of times the participant ate “other meat besides pork” in the last week, as indicated in Sections 1A and 2A of the attached survey. In our analysis, we will measure the longer-term effect of watching the video using the difference between self-reported total meat consumption in the follow-up online survey (Section 2A) and self-reported total meat consumption before watching the video (Section 1A).

We also plan to assess the (immediate) effect of our intervention on various emotional outcomes. These measures are based on participants’ responses to the six questions in Section 1F of the attached survey. For each of these questions, participants’ responses will be expressed on a 0 to 3 scale, with 0 representing no effect on the indicated emotion and 3 representing a very large effect.

Lastly, we consider the degree to which participants avoid the information presented in our treatment video. Our main measure of information avoidance is based on a choice, presented after 11 minutes of watching the assigned video, of whether to stop watching (i.e., to skip the last 5 minutes) or to continue watching the video. This choice is shown in Section 1D of the attached survey file. Our survey also includes video comprehension questions, a self-reported learning question, and a self-reported attention question (see Sections 1E and 1H of the attached survey) that may serve as alternate measures reflecting the degree to which participants received the information presented in their assigned video.

We will be conducting a randomized controlled trial, as described in the Interventions section above.

Not available

Qualtrics built-in randomizer tool.

Individual.

No

Individual-level intervention.

1,200 individuals (estimated), with the exact number depending on the number of participants allocated by the Rotman Behavioral Research Lab from the Rotman Credit Pool, which is a sample of undergraduate students at the Rotman School of Management who agree to participate in behavioural research studies in exchange for course credit. We have been allocated 600 participants in the Fall Semester of 2024 while our allocation in the Winter Semester of 2025 has not yet been determined. Thus, our estimated total sample size of 1,200 is based on an assumption that our winter allocation will be the same as our fall allocation.

Half of our sample will be assigned to the treatment condition, and the other half will be assigned to the control condition.

Assuming that 70% of participants choose an item with meat in their first menu choice (before watching a video), and assuming that 10% of participants choose a non-meat item in their first menu choice and then switch to a meat item in a subsequent menu choice, then a sample size of 906 is needed to measure a 5 percentage point effect size with 95% confidence and 90% precision, using a paired t-test to test for a significant effect among treatment group participants. If 25 percent of participants invited to participate fail to complete both parts of our study (for example, by not showing up to the initial in-person session or by not completing the follow-up online survey sent seven days after the in-person session), then the target sample size would become 906/0.75 = 1,208 participants. Note that additional statistical power may be needed to differentiate a treatment group effect from a potential control group effect.