Latest / The Vegan Report / Nothing Seems to Work | Science of Activism
Transcript
- 0:00Hi, my name is Seth Ariel Green and I'm glad
- 0:04to be here on The Vegan Report. I'm a researcher
- 0:07at the Humane and Sustainable Food Lab at Stanford
- 0:09University. We're a research group at the medical
- 0:13school there aimed at accelerating society's
- 0:15transition away from factory farming. Today we're
- 0:19looking at what gets people to cut back on consumption
- 0:22of meat and animal products, specifically persuasion,
- 0:26choice architecture, psychological appeals, and
- 0:30adding new plant -based meat analogues to menus.
- 0:33I'll be discussing two papers that I'm a co -author
- 0:35on. The first is called, Meaningfully Reducing
- 0:39Consumption of Meat and Animal Products is an
- 0:41Unsolved Problem, a Meta -Analysis. The second
- 0:45is a preprint called, Taking a Bite Out of Meat
- 0:48or Just Giving Fresh Veggies the Boot. Plant
- 0:50-based meats did not reduce meat purchasing in
- 0:52a randomized controlled menu intervention. If
- 0:55you're interested in this work, you can check
- 0:57out the lab I work at at foodlabstanford .com.
- 1:02That's foodlabstanford .com. We have all of our
- 1:06publications there, including the two papers
- 1:08that I'm about to discuss. And I also write a
- 1:11sub -stack called regression to the meat. That's
- 1:14regression to the M -E -A -T. The lab is run
- 1:18by Maya Mathur, who is a professor at Stanford.
- 1:21And I am a research scientist there. You can
- 1:24also find new episodes every Wednesday on this
- 1:26podcast. Okay, so we're going to go as follows.
- 1:30The meta -analysis, the first paper, is the main
- 1:32paper I'm going to be talking about. But because
- 1:35meta -analysis is a whole complicated thing,
- 1:37I'm going to describe the method in general and
- 1:42like why one would do this kind of paper. And
- 1:46then we're going to look at the paper itself.
- 1:48Next, I'll turn to the second paper. I'll start
- 1:50with its setup, and I'll then move to its results.
- 1:54And finally, we'll conclude with general lessons
- 1:56from the two. So that's five things I hope to
- 1:58cover. Maybe six. Okay, so let's start out with
- 2:03why meta -analysis, or rather, what is the whole
- 2:06point of this thing? Basically, a meta -analysis
- 2:08is a paper where you take the results of other
- 2:10papers and you amalgamate them into a conclusion
- 2:14about a literature as a whole. So, Let's say
- 2:17like in the idealized case, you have 10 papers
- 2:21that all look at one thing. Let's say it's the
- 2:24effects of insecticide treated malaria nets on
- 2:29infant mortality. If you have those 10 papers,
- 2:32meta -analysis is pretty easy. Basically what
- 2:34you do is you put them all together, you look
- 2:39at the outcomes, and then you come up with a
- 2:42weighted average of those outcomes. where more
- 2:47weight goes to more precise estimates. And in
- 2:50practice what this means is that bigger papers,
- 2:53those with more subjects, get more weight in
- 2:56the resulting average. Okay so that's the idealized
- 2:58case. Like the intervention is one thing and
- 3:03the outcome is one well -defined thing and you
- 3:06just give a weighted average of all the 10 interventions
- 3:10on the one thing and see like what the overall
- 3:12effect is. And the idea here is that If you look
- 3:16at 10 studies from 10 different locations, you're
- 3:19going to have a better sense of what the so -called
- 3:22true underlying population effect is than you
- 3:25would if you just chose one study. I think this
- 3:29is pretty intuitive. It's the same way that,
- 3:32like, if you want to figure out what Americans
- 3:33think about politics, you do a random sample
- 3:35of 100 people, not just 10 people or not just
- 3:38one person. Like, you want just more samples,
- 3:43or a larger sample rather. Okay, that's like
- 3:46the best -case scenario My lab and well my work
- 3:51more generally we're in the behavioral sciences
- 3:54and Meta analysis can be a little trickier than
- 3:58that The first is that like let's say you wanted
- 4:02to ask a question. This is the question that
- 4:04my paper actually asks What gets people to eat?
- 4:09less meat and animal products okay, so It's pretty
- 4:16clear what the outcome is. We're going to look
- 4:17at reductions in meat and animal products, which
- 4:20I'm henceforth just going to call MAP. So MAP
- 4:22is meat and animal products. But it's not at
- 4:28all clear that the research field as a whole
- 4:30agrees on what the right way to approach that
- 4:32is. So you could theoretically write this paper
- 4:35and look at thousands of different approaches.
- 4:41And in that case, it's not totally clear what
- 4:46any weighted average you come up with actually
- 4:49means. So like let's say we're looking at two
- 4:53papers. We want to meta -analyze them and one
- 4:58is a choice architecture intervention. That's
- 5:00the kind of thing where you manipulate something
- 5:02about an environment where someone eats meat
- 5:04and then sees see how that affects how much meat
- 5:09someone eats, right? So that could be a good
- 5:11example is like making the spoons with which
- 5:14people at cafeterias serve meat literally smaller,
- 5:18or like putting the plant -based option at eye
- 5:22level to try to guide people unconsciously towards
- 5:24that. Okay, so you have one study that's like
- 5:26that, and the other study is you make people
- 5:28watch Dominion, which is this really upsetting
- 5:31documentary of how animals are treated on factory
- 5:36farms. Okay, and then you average those two things
- 5:38together, and you get a number, but like, What
- 5:42does that number mean? What is the average of
- 5:47changing the spoon size and watching Dominion?
- 5:50In some sense, it could be said to represent
- 5:53the literature's best efforts. Like if you said,
- 5:57this is what our field is doing, this is how
- 5:59we're doing, this is our number. But in another
- 6:02sense, that average doesn't really correspond
- 6:05to anything you can... point to in the real world
- 6:08because these two things are just not alike.
- 6:10You cannot average the apple -ness of an apple
- 6:15and an orange. And then the other side of this
- 6:18problem is that the same thing is true of outcomes.
- 6:21So if one study reports food frequency questionnaires,
- 6:25which are just, you're supposed to tell a researcher
- 6:28what you ate over the past 24 hours, and the
- 6:32other is watching what people do and don't eat
- 6:38at a dining hall, it's also not clear what the
- 6:41average of these two outcomes actually signifies.
- 6:44Okay, so this is why meta -analysis is sometimes
- 6:47taken as like the definitive answer on what is
- 6:50the effect of X on Y, but when you read these
- 6:52papers, the hard question to ask is like, are
- 6:58these inputs even similar enough that we think
- 7:02they can be averaged, like conceptually? And
- 7:06likewise, are these outcomes all pointing at
- 7:10the same target? And if the answer to those questions
- 7:13is I'm not sure or I don't know, then you can
- 7:19come up with an answer to your question, but
- 7:22it's just not clear that it means anything. Okay,
- 7:26so that's why meta -analysis is hard. Nevertheless,
- 7:31I did one with my co -authors Benny Smith and
- 7:34Maya Mathur. on the question of what gets people
- 7:38to eat less meat and animal products map, or
- 7:42which class of interventions is most effective.
- 7:47The way that we handle this problem of, wow,
- 7:50studies are so crazily diverse and we can't really
- 7:52make sense of everything, is like so. First,
- 7:57we only looked at studies that met a pretty high
- 8:00bar for research validity. And that meant a few
- 8:05different things. First, they actually had to
- 8:10measure consumption, not behavioral intentions
- 8:14or not attitudes. Pausing here to say they can
- 8:17also measure behavioral intentions and attitudes,
- 8:20but they had to measure consumption behavior,
- 8:22and we only looked at that outcome. So at least
- 8:24we're comparing like to like basically for the
- 8:29results we're looking at within papers. And the
- 8:32second criteria was that they needed to have
- 8:35enough people in their samples that we thought
- 8:38that we could say something meaningful. So that
- 8:40was at least 25 people in treatment and 25 in
- 8:43control. And okay, I've got a caveat here, something
- 8:49complicated. Not every intervention is assigned
- 8:55at the level of individual. Some things are like
- 8:58on some days cafeterias do something and then
- 9:01not on other days And it's not totally clear
- 9:03how many people came in and out So for those
- 9:06kinds of studies studies where treatment was
- 9:08administered at the level of day or place We
- 9:12just we had to say we said at least five Units
- 9:16in treatment and control and typically in all
- 9:18of those cases. There were like hundreds of people
- 9:20actually coming through Okay Third criteria it
- 9:26had to be a randomized control trial not any
- 9:28other kinds of design. This is just like for
- 9:31Getting causal estimates that is like X caused
- 9:37Y RCTs Trump other designs and Fourth they had
- 9:43to measure outcomes at least a single day after
- 9:48Treatment was first administered and the idea
- 9:51here was You might have an intervention that
- 9:56really really effectively reduces meat for like
- 9:58one meal I mean theoretically you might just
- 10:01like not serve But you might make it like really
- 10:04hard or difficult to get meat like what if you
- 10:07had a what if you went into a restaurant? And
- 10:09if you ordered a burger, they gave you an electric
- 10:11shock and you would be like, oh my god Why'd
- 10:13you do that? But that would definitely have an
- 10:16effect but then like what happens when you go
- 10:18home you probably Hate the people who did this
- 10:22to you and eat like twice as many burgers. So
- 10:25we only look at studies that measure a delayed
- 10:29effect of some sort to try to get at like how
- 10:32do people deal with the intervention in their
- 10:34lives, like how do they compensate later. Okay
- 10:37so that's the setup and when we did that we did
- 10:44this like long thorough literature review and
- 10:47we came up with about uh that 35 papers met all
- 10:52our criteria and that was 41 interventions and
- 10:55a hundred and 41 sorry studies 112 interventions
- 10:59so this literature when we first started looking
- 11:02at it it's like thousands or maybe tens of thousands
- 11:05of papers i mean broadly speaking not all of
- 11:08them look strictly at like reductions in meat
- 11:10and animal products but some of them might look
- 11:12at like you know trying to get kids to drink
- 11:15like oat milk or even like you know, just cut
- 11:19back on fatty foods in general. It's a lot of
- 11:21different things here, but it started out as
- 11:24a huge literature, ended up with 35 papers. Most
- 11:29of those papers have been published since 2020,
- 11:32so this is a pretty modern literature. Okay,
- 11:35so 35 papers that meet all our criteria, which
- 11:38again are about quality, and they actually have
- 11:39to be trying to get people to eat less meat and
- 11:43or animal products, and those papers fell into
- 11:46four categories. Those were persuasion, choice
- 11:50architecture, psychology, and then these kind
- 11:53of combo persuasion and psychology papers. Going
- 11:56through each in turn. Persuasion is what it sounds
- 12:00like, is trying to get people to eat less meat
- 12:02by talking to them about it, or otherwise using
- 12:04information to consciously persuade them. Persuasion
- 12:09efforts can come in three broad categories, broadly
- 12:12speaking, and those are, I bet you can see this
- 12:14coming, health, the environment, and animal welfare.
- 12:18So an environmental intervention, a good example
- 12:21is a paper by Andrew Jalil and others from 2023
- 12:25that replaced an intro economics course with
- 12:29just one lecture of it with a lecture about the
- 12:34global warming consequences of meat. It also
- 12:37covered a little bit about the health into stuff.
- 12:40The health version of this is aimed at mostly
- 12:45talking about red and processed meat and how
- 12:48that's not good for you. But there are some health
- 12:50interventions that just broadly make the point
- 12:52that, like, you should be eating more vegetables.
- 12:55So we looked at those. And then the last one,
- 12:57the animal welfare stuff, this is like pamphlets
- 13:00and videos and op -eds. A lot of these studies,
- 13:05the animal welfare stuff, was run by animal welfare
- 13:08organizations. Okay. Those are the persuasion
- 13:12approaches. Then there is the choice architecture,
- 13:14which is, I've talked a little bit about earlier,
- 13:16but to rehash, that's changing the aspects of
- 13:19the context in which food is selected and consumed.
- 13:22So a good example, Anderson and Nelander, 2021,
- 13:26they, as I said earlier, just put the vegetarian
- 13:32option at a university cafe for the day at eye
- 13:35level or not. So that just changes how you see
- 13:38it or not. And then psychology studies are really
- 13:42interesting broad category here basically they
- 13:48want to manipulate how you think and feel about
- 13:53meat without necessarily giving you like direct
- 13:55reasons that you should eat it so it's not like
- 13:58meat is bad for you so you should eat less meat
- 14:00it's more like oh did you know that plant -based
- 14:03options are increasingly popular and cool among
- 14:07your peers that last example that's like increasingly
- 14:10popular and cool this is the norms approach.
- 14:13And most psychology studies we looked at were
- 14:16norms -based interventions and those basically
- 14:19try to make plant -based options seem like either
- 14:23normal or desirable. Those are two different
- 14:26kinds of norms interventions. And further, you
- 14:29might say that plant -based options are normal
- 14:32today or you might do this other thing called
- 14:34dynamic norms where you say plant -based options
- 14:39are increasing in popularity. And the idea is
- 14:42there that people will do this pre -conformity
- 14:44thing where they want to go to where people are
- 14:48going in the future. And finally, we have a couple
- 14:52of papers that combine persuasion and psychological
- 14:56messages. One example is from Piester et al.
- 15:02P -I -E -S -T -E -R 2020. They put up messages
- 15:07around a dining hall, which I'm pretty sure was
- 15:09UVA, but it's like some school in South Central
- 15:12America, or the central region of the South in
- 15:17America. And they say, veggie burgers are a tasty,
- 15:23healthy choice. they're good for the environment,
- 15:27they're sustainable. Also, 95 % of your peers
- 15:30agree that veggie burgers are the tasty choice.
- 15:32So that's doing some things to try to tell you
- 15:35why you should eat them, and it's doing some
- 15:37things that are trying to tell you how you should
- 15:39think about or feel about the broader question
- 15:41of eating plant -based options. Okay, so those
- 15:47are our big approaches. We've got persuasion,
- 15:51choice architecture, psychology, and combo of
- 15:54persuasion and psychology. What are our results?
- 15:58Doing a little drumroll here. Overall we find
- 16:02nothing very impressive. In meta -analysis it's
- 16:06common to standardize everything by turning it
- 16:09into hundredths of a standard deviation, so it's
- 16:14called Cohen's d, and a d of 1 is one standard
- 16:19deviation of change. A way to think about this
- 16:21is Like, if I did an intervention on children's
- 16:25SAT scores, and I achieve the coincity of 1,
- 16:29that would be equivalent to 100 points each on
- 16:31the verbal and math increase. Because the standard
- 16:36deviation of the SAT is 100 points per section.
- 16:39In this context, it's a little less clear what
- 16:41a standard deviation means, but like, just go
- 16:43with it. It means like, a D of 1 is like, oh,
- 16:46you really got the job done. Our overall... Cohen's
- 16:52d is 0 .07. That means 7 one hundredths of a
- 16:57standard deviation. In general, this corresponds
- 17:00to like a few percentage points change or up
- 17:03to like five percentage points in some cases,
- 17:06or maybe like a, you know, notable reduction
- 17:09that levels out over time. Basically, we put
- 17:13everything together and we do not come up with
- 17:15the kind of transformational changes that we
- 17:17were looking for. That's why the paper is titled
- 17:20meaningfully reducing meat consumption is an
- 17:22unsolved problem. Now let me harken back to the
- 17:29problem I was saying earlier about how it's not
- 17:31totally clear that when you put all these studies
- 17:33together that it means anything because I don't
- 17:35know what the average of a psychology study at
- 17:37a choice architecture study actually indicates.
- 17:40To deal with that we then separated each approach
- 17:44into its own theoretical category and we looked
- 17:47at those numbers by themselves. Basically here's
- 17:49what we found. Choice architecture studies have
- 17:52the largest effect on average, about 0 .2, which
- 17:55is again about a fifth of a standard deviation
- 17:57of change. And that's, you know, in some contexts
- 17:59that could be meaningful. But we only had two
- 18:02studies that met that. Meaning that almost every
- 18:06choice architecture study, like hundreds of them,
- 18:08did not qualify. Typically that's because a lot
- 18:11of these choice architecture things either manipulate
- 18:14hypothetical, environments though it'll like
- 18:17take you through an online grocery store and
- 18:20then like you know make the Plant -based option
- 18:22more prominent, but you don't actually eat anything
- 18:24Or choose anything and the other reason is that
- 18:27everything is very immediate. So a good example
- 18:29is in the defaults literature It's a good example
- 18:35The paradigmatic defaults intervention would
- 18:41be let's say you were hosting an academic conference
- 18:47and you were asked in advance what kind of meal
- 18:49you wanted and your choices were vegetarian and
- 18:53non -vegetarian. In the default intervention
- 18:57version, your choice would just say, you will
- 19:01be getting a vegetarian meal. Please email us
- 19:03if you don't want that. That's great. It seems
- 19:06to have worked very well. My co -author, Maya,
- 19:08has a couple of really strong papers in this
- 19:09category, but we don't know what people do when
- 19:13they go home. or when they like leave the conference.
- 19:16So yes, there's like some evidence that just
- 19:20making it a little annoying to get meat, like
- 19:22you have to write someone an email to say, yeah,
- 19:24I want the beef, reduces the amount of meat people
- 19:27order. But whether that has any lasting effect
- 19:30on people's behavior, I don't know. Okay, so
- 19:35that's choice architecture. I would say psychology
- 19:38is the second most or psychological interventions.
- 19:42They have like about a tenth of a standard deviation
- 19:44of change. This dynamic norm stuff, if you want
- 19:47to read more about it, there's lots of cool papers.
- 19:49If you search like dynamic norms that meet, you'll
- 19:51probably encounter papers by Greg Sparkman, who's
- 19:55a professor at BC. I would say his lab is like
- 19:58the main center of this kind of research. It's
- 20:02really hard to use these interventions to get
- 20:04people to do things. Or you can get them to change
- 20:07a little bit, but it's like it works at a classic
- 20:10paper with Greg and co -authors. finds good changes
- 20:13in one setting, no changes in another, and like
- 20:16a backlash in the third. And when you look at
- 20:19this all together, it's really unclear why it
- 20:23worked in one place and not in another. We don't
- 20:26know. Like one explanation is that maybe there's
- 20:29something going on that we don't know about why
- 20:31it works and better in some context and others,
- 20:33not others, but the other possibility is that
- 20:35it's just really noisy. Meaning like the true
- 20:39answer is zero, or pretty close to zero, but
- 20:44there's spread around that just because of random
- 20:47variation, so it's going to look sometimes like
- 20:51you get a positive result, but it's just noise.
- 20:54Okay, but I don't really know. That's sort of
- 20:56an ongoing area of research. In general, persuasion
- 21:00efforts, we get a D of 0 .07, which is the same
- 21:03as our overall, and that masks a little bit of
- 21:06heterogeneity. In general, animal welfare appeals
- 21:09don't seem to be working. We get a D of 0 .03
- 21:14for all animal welfare appeals. That's not enough
- 21:18to do anything with. It's 0 .09 for the environment
- 21:22and 0 .08 for health. These are, again, all I'm
- 21:25just saying about these is that they're really
- 21:26small numbers. For animal welfare, not even a
- 21:31percentage point change on average. And for health
- 21:33and environment, maybe a couple percentage points
- 21:35changes on average. And persuasion and psychology
- 21:39combos are like a D of .11, which is about as
- 21:45effective as psychology on its own. Okay, so
- 21:49what do we make of this? I still don't know how
- 21:58to digest these results in, I wrote this stupid
- 22:01paper, this great paper. Do we just stop? I mean,
- 22:08that's very discouraging, but I would say that
- 22:12it's discouraging to me that if you look at like,
- 22:15there have been some previous meta -analyses
- 22:17that found big results and what we think basically
- 22:22happened, or I'm just going to say me, what I
- 22:24think happened, is that those papers hoovered
- 22:27in studies with much weaker designs and measurement
- 22:31strategies and weaker designs are just more likely
- 22:36to produce stronger effects for complicated reasons.
- 22:41This is what the online writer Guern, G -W -E
- 22:45-R -N, calls the stainless steel law of evaluation,
- 22:48which is that the better designed the impact
- 22:50assessment of a social program, the more likely
- 22:52is the resulting estimate of net impact to be
- 22:55zero. So we only look at like what I consider
- 22:58pretty good studies and we found basically no
- 23:01effect. That's, I mean, so in some ways it's
- 23:03That's discouraging. On the other hand, I would
- 23:07say this literature is getting better and better,
- 23:10like in terms of its methodological quality.
- 23:12As I said, most of the papers we looked at are
- 23:14from pretty recently. That's great. We still
- 23:16think it's a lot of untapped ground. And also
- 23:19there's a lot of theoretical stuff that can be
- 23:21done that isn't in our data set. So for instance,
- 23:27what about like extended contact with farm animals?
- 23:29Does that change anything? manipulations to the
- 23:32price of meat itself, like little stimulated
- 23:35small -scale meat taxes. None of those studies
- 23:38have been tested in a way that meets our criteria.
- 23:42There was a dearth of disgust studies, like,
- 23:45ew, oh my god, how much... I had no idea that
- 23:48so much milk and blood and pus was allowed in
- 23:51milk. Almost nothing that evaluated that was
- 23:54done rigorously. I think there's some pretty
- 23:57good potential for fiction to be rigorously evaluated,
- 23:59like... go ahead and show a lot of people like
- 24:02Lisa the Vegetarian, which is a Simpsons episode,
- 24:05or Babe, which made its farmer, the guy who played
- 24:09the farmer, James Cromwell, he went vegan after
- 24:11being in that movie, or Okja, which is like explicitly
- 24:15an animal rights parable. So I still think there's
- 24:18a lot of room here to go. Now, the next thing
- 24:23is that some people might say, what about like
- 24:28fake meat options? Isn't that the thing? Doesn't
- 24:30everybody love Beyond Meat or Impossible Meat?
- 24:33You haven't talked about that. Why not? Which
- 24:36gets us to our next paper, which, again, is called
- 24:40Taking a Bite Out of Meat or Just Giving Fresh
- 24:43Veggies the Boot. Plant -based meals do not reduce
- 24:46meat purchasing in a randomized, controlled menu
- 24:48intervention. Now, that title basically tells
- 24:51you what the whole paper is about, but I'll give
- 24:53you the setup. So it's an online survey. We get
- 24:58people into the survey and we give them questions
- 25:02about like choosing pens, choosing t -shirts.
- 25:05It's like, you know, we tell them it's a consumer
- 25:07choice survey. And then one of those things is
- 25:10they choose a taco filling from a menu that looks
- 25:14a lot like Chipotle's. And they were randomized
- 25:17into three treatments. The first is the only
- 25:21plant -based option is veggies and guac. and
- 25:24you got a bunch of meat options like beef barbacoza
- 25:27or barbacoa i think and like carnitas and chicken
- 25:31and maybe and steak i think okay that's the first
- 25:34treatment arm the second is you got the veggies
- 25:40and guac but you also have sofritas which is
- 25:42this thing that chipotle has it's like a you
- 25:45know pork analog tofu based thing if you've been
- 25:48to chipotle i think it's pretty good and the
- 25:50third option is this hypothetical thing we invented
- 25:54and showed people called chickenetus, which is
- 25:56like a fake chicken plant -based analog. And
- 25:59the idea is just seeing more plant -based options,
- 26:04or does having more plant -based options available
- 26:06make you more likely to select plant -based?
- 26:09Uh, the answer is kind of, like a little, but
- 26:14not much and not enough to be statistically significant.
- 26:18adding sofritas reduced demand for meat by 1
- 26:21.14 percentage points relative to having just
- 26:25chips and guac or veggies and guac and adding
- 26:29sofritas and chickenitas together reduced demand
- 26:31for meat by 2 .14 percentage points so you know
- 26:38two percentage points is something if we could
- 26:41like do that at all the Chipotle stores in America
- 26:43that would probably be a big something But we
- 26:45were aiming for five percentage points change,
- 26:48which we think is basically like the smallest
- 26:50amount we could feasibly tell Chipotle, hey,
- 26:53look, there's unmet demand for plant -based options.
- 26:56So you should add another one. We also noted
- 27:02that as we added more and more veggie options,
- 27:07demand for chip for the veggies and guac options
- 27:10decreased. So it started out. in the no no plant
- 27:18-based meat analogs options it was 9 .2 percent
- 27:23and then in the arm with sofritas it goes to
- 27:256 .6 percent and the arm with sofritas and chickenitas
- 27:28it goes to 5 .7 percent so one unfortunate conclusion
- 27:32to this is that when you add a new plant -based
- 27:35meat option meat analog option to your menu at
- 27:39least some of that demand is coming from vegetarians
- 27:42or people who would otherwise be eating vegetarian.
- 27:50The paper has some other results that are kind
- 27:52of like just confirming that this is basically
- 27:54legit. For instance, men eat more meat than women.
- 27:58Republicans eat more meat than Democrats. People
- 28:01without a four -year college degree eat more
- 28:04meat than people who do or choose more meat in
- 28:07this context. When I put all these results together,
- 28:13it sure might sound like I'm a huge hater. Like
- 28:17I'm saying, oh, nothing works, sorry, go home.
- 28:22Sometimes I do feel that way. This is a really
- 28:24tough problem we're working on. I think that's
- 28:26overall the conclusion that I would give from
- 28:30these papers is that it's really hard to get
- 28:33people to eat less meat and animal products.
- 28:36Anyone who has told you otherwise, my honest
- 28:39-to -god opinion, is that they probably weren't
- 28:44evaluating the intervention very rigorously.
- 28:47And I don't, that's like nothing special about
- 28:50our field, this is just true about social science
- 28:53research. Again, the better design and evaluation,
- 28:57the more likely the resulting estimate of net
- 29:00impact is to be zero. On the other hand, I still
- 29:03think there are lots of cool things that should
- 29:04be tried and can be tested rigorously. And what
- 29:07I'm personally interested at the moment is something
- 29:09closer to like Mental categorization so like
- 29:12how do vegetarians think about the question of
- 29:14eating meat versus eating eggs things like that
- 29:17and Lastly I would say If and when plant -based
- 29:25meat is as good as or even cheaper than Meat
- 29:32that's gonna be a game changer, but I just wish
- 29:36to caveat that by saying that There does not
- 29:39seem to be any evidence of that happening anytime
- 29:41soon. So, again, sorry to be the bearer of bad
- 29:46news. What I'm saying is basically that this
- 29:48is a hard problem, and we just need to keep cracking
- 29:53at it. Thanks for listening to my bucket of cold
- 29:57water of a podcast recording. If you have any
- 30:02questions, you can find me at SethArielGreen
- 30:05.com. Green like the color. You can find our
- 30:09lab at foodlabstanford .com. You can find my
- 30:13newsletter at regressiontothemeat .substack .com.
- 30:19Thanks very much for listening and I look forward
- 30:21to hearing from you if you have any questions.
- 30:23Thanks.