Yep, It Replicates! Students Eat Less Meat After Studying the Ethics of Factory Farming

In 2020, Brad Cokelet, Peter Singer, and I published a study finding that U.C. Riverside students eat less meat after studying ethical issues concerning meat.  Over 1000 students in large lower-division philosophy classes were assigned either a short article, a 50-minute discussion meeting, and an optional video on meat ethics (focused on factory farming) or similar materials on charitable giving.  We then looked at their food purchases at selected campus dining locations by examining anonymized purchase receipts.

Percentage of purchases containing meat, 2020 study:

Meat ethics condition: 28% before intervention vs. 25% after (p = .004)

Charitable giving control condition: 28% before vs. 29% after (p = .55)

Looking only at purchases of $4.99 or more (likelier to be full meal purchases, but also including expensive snack and drink purchases), the numbers declined from 52% to 45% in the meat ethics condition (p = .001) while staying steady at 52% in the control condition.  The estimated effect duration is several weeks to a few months.

It was a striking result!  Although the effect was small, I had expected no effect at all, partly because ethics professors don't appear to behave any differently than other groups of professors, including specifically on the issue of vegetarianism: Ethics professors in the U.S. are much more likely than other professors to say it's morally bad to eat meat while they self-report eating meat at about the same rate as other professors.  If professional ethicists don't behave any differently as a result of decades of exposure to philosophical ethics, why should we expect a brief intervention on students to have any effect?

I confess I was nervous about whether the effect was some sort of fluke, despite its statistical significance and the care with which we conducted the study.  Would it really replicate?

Yes, it does, as Brad, Peter, and I find in a new paper published last weekend.

Our new study differed from the 2020 study in a few important ways (explained below), but the main intervention was basically the same (a reading, a 50-minute discussion meeting, and an optional video), and the dependent measure was again meat purchases at U.C. Riverside campus dining locations.  We found:

Percentage of purchases containing meat, 2021 study:

Meat ethics condition: 30% before intervention vs. 23% after (p < .001)

Comparison group: 31% before vs. 30% after (p = .79)

Looking only at purchases of $4.99 or more, the numbers declined from 51% to 42% in the meat ethics condition (p = .001) while staying steady at 52% to 53% in the comparison group.

We also found that students (anonymously) expressed more agreement with the statement "Eating the meat of factory farmed animals is unethical" after the intervention than before (37% vs. 54% agreeing, p < .001).  Also, at the end of the discussion meeting, students were given the opportunity to anonymously pledge to avoid eating factory farmed meat for the next 24 hours.  Forty-three percent did so (in a student population estimated to be < 5% vegetarian) and 76% of those who pledged later reported having followed through on the pledge.

There are some differences between the 2020 and 2021 studies that strengthen our findings, beyond simple replication.

First, in the 2020 version students were invited to view the optional vegetarianism advocacy video at home, on their own time.  About half reported having watched at least some of it.  This raises the worry that the film might have been driving the effect.  Maybe the more traditional aspects of instruction (the reading and discussion) would have had no effect in isolation?  In the 2021 version, instead of using an opt-in structure we used an opt-out structure and we split participants into film and non-film conditions.  Half of the students were not told about the video (the non-film condition).  For the other half (the film condition), the video was shown at the beginning of their discussion meeting and students were invited to step outside for 11 minutes if they preferred not to watch it.  (Only one student out of 358 in this condition actually did so, though we discovered after the fact that some students "opted out" by closing their eyes during some of the footage.)  The purchases of students in both the film and the non-film conditions were then compared with the purchases of a group of other U.C. Riverside students who were not enrolled in the target courses.

Second, in the 2020 version all of the teaching assistants leading discussion were vegetarian.  This raises the concern that the teaching assistants might have been biased in their presentation or that the effect might have arisen through the social influence of teaching assistants, who might have implicitly or explicitly revealed their personal behavior.  In the 2021 version, the majority of instructors were not vegetarian.

So... was there a difference between the film and non-film conditions?  We could find no statistically detectable difference between conditions either in expressed attitude (among 573 respondents) or in pledge rate (among 751 attendees).  However, it is possible that the effect on actual meat purchase behavior was larger for students in the film condition:

Percentage of purchases containing meat, 2021 study, by condition:

Film condition: 30% before intervention vs. 21% after (p < .001)

Non-film condition: 30% before vs. 25% after (p = .047)

Was there a difference between students of the vegetarian versus non-vegetarian teaching assistants?  Here were found no statistically detectable difference in attitudes, pledge rates, or purchase behavior.  For example,

Percentage of purchases containing meat, 2021 study, by instructor attitude:

Vegetarian instructor: 30% before intervention vs. 23% after (p = .010)

Non-vegetarian instructor: 28% before vs. 23% after (p = .014)

Thus, it appears that neither showing the film nor having a vegetarian instructor is necessary for the observed effect.

I am also encouraged by seeing some similar results at Occidental College, where students also appeared to purchase less meat after a class meeting on the climate and health benefits of a vegetarian diet.

Full details of the study here (including links to the pre-registration and the raw data).

[image source]