Intervention(s)
We will randomly assign each participant to one of six conditions. These conditions are the treatment groups A-D (described below) plus a control condition and a placebo condition.
Participants in the control condition will read text that describes the role of food in everyday life and no additional text. Participants in conditions A-D will read additional text that will be specific to the condition. The subjects in the placebo group will read a text unrelated to food; specifically, it will be a text about the importance of breathing clean air.
Additional text participants will read in conditions A-D:
A. Provide subjects with (factual) “social information” about dietary choices of the population, for example “thousands of people are adopting a plant-based diet”.
B. Put subjects in a “loss domain” state, with text that cites the average number of animals that die each year as a result of an individual eating a non-plant-based diet.
C. Put subjects in a “gain domain” state, with text that cites the average number of animals’ lives that an individual can save each year by adopting a plant-based diet.
D. Sentience/sapience/morality: Highlight animals’ ability to feel pain, and the fact that many standard farming practices inflict pain.
Next we will assess whether the provision of specific information affects opinions about eating animal-derived products. Within a set of general questions about attitudes, we will ask participants to indicate if a sentence like “If the taste and texture were identical to beef, I would be likely to eat a vegetarian burger” applies to them. This simple design will provide us with immediate self-reported information about whether attitudes change in the different conditions. The advantage of simplicity comes with the cost that the results would be based on “stated” rather than “revealed” preferences. Given also the nature of the question of interest, demand effects and social desirability biases are possible. This concern, however, is unlikely to be severe in our case because we are interested not only in the difference between the treatments and the control, but also in the differences between each of the individual treatments. To the extent that demand effects and social desirability biases are present in all conditions, looking at the differences would help “net out” this potential confound.
In addition to the above-described research questions, we will have all study participants complete the “Big Five” personality markers (Gosling, S. D., Rentfrow, P. J., & Swann, W. B., Jr. (2003). A Very Brief Measure of the Big Five Personality Domains. Journal of Research in Personality, 37, 504-528). Including these types of questions will help us determine whether key variables of interest regarding changes in dietary choice correlate with personality measures.
The main hypothesis that we are testing is whether the behavioral interventions and/or the provision of morality-based information leads to more aversion toward eating animal products. However, we may find different results. For example, views and habits with respect to diet tend to be stable and, we may find participants are impervious to information focused on eliciting moral concerns. Alternately, individuals may even resist information which conflicts with their baseline belief that their diet is morally acceptable, and perhaps cognitive dissonance or some other mechanism may cause them to become further entrenched in their current beliefs and behavior.
To test our hypotheses, we will examine whether there are differences across conditions in the attitude and behavior measures we include in the study.
To screen out invalid responses, we will have 3 coders go through participants' summaries of the message they just read and indicate (yes/no) whether they believe the participant has shown evidence that he/she read the assigned message. If all 3 coders unanimously agree that a participant has not shown evidence of having read the assigned message, then that participant will be removed. Otherwise, the participant will be included in all analyses.