Research report: adding a plant-based option (or two) doesn't change behavior

By Seth Ariel Green šŸ”ø, Jessica Hope, Jacob_Peacock, MMathuršŸ”ø @ 2025-09-08T14:02 (+64)

This post summarizes a new preprint from the Humane and Sustainable Food Lab at Stanford titled ā€œTaking a bite out of meat, or just giving fresh veggies the boot? Plant-based meats did not reduce meat purchasing in a randomized controlled menu intervention.ā€ 

The paper reports on an online RCT where participants were asked to select taco fillings from one of three menus designed to mimic the options at Chipotle. They could choose from both meat- and plant-based options.

The randomized treatment was whether participants saw

The treatment was embedded in a series of decoy questions about choosing pens and t-shirts to obscure the purpose of the study.

The main outcome was whether people chose a meat-based taco filling or not. Our second outcome was whether Chickā€™nitas reduced demand for chicken specifically among meat-based options. All outcomes were hypothetical, and price was held constant.

4,431 people selected a taco filling.

We powered the study to detect a 5 percentage point (pp.) reduction in meat selection, the smallest effect size which we thought might prompt Chipotle or someplace similar to consider adding a new PMA.

Unfortunately, we didnā€™t find an effect of that magnitude. 

Compared to having just veggies:

Adding Chickā€™nitas modestly but insignificantly decreased demand for chicken, but simultaneously, demand for steak increased. Going from no PMAs to two decreased demand for veggies & guac from 9.2% to 5.7%. 

See table below for full breakdown. 

 

No-PMA Arm

(n = 1,501)

Status Quo Arm

(n = 1,479)

Chickā€™nitas Arm

(n = 1,451)

Chicken

613 (40.8%)

580 (39.2%)

516 (35.6%)

Steak

412 (27.4%)

411 (27.8%)

434 (29.9%)

Beef Barbacoa

184 (12.3%)

185 (12.5%)

182 (12.5%)

Carnitas

129 (8.6%)

138 (9.3%)

139 (9.6%)

Veggie

138 (9.2%)

98 (6.6%)

83 (5.7%)

Sofritas

n/a

56 (3.8%)

40 (2.8%)

Chickā€™nitas

n/a

n/a

42 (2.9%)

Declined to order

25 (1.66%)

11 (0.74%)

15 (1.0%)

Meat

1,338 (89.14%)

1,314 (88.8%)

1,271 (87.6%)

Non-meat

138 (9.19%)

154 (10.4%)

165 (11.4%)

Men, Republicans, people without a four-year college degree, and people who ate more than ten or more servings of meat per week were all more likely to select a meat-based taco filling than folks in other categories.

We donā€™t know if weā€™d have found bigger effects in stores. Our experiment didnā€™t include any sensory information, so perhaps the Chickā€™nitas would have been more appealing in person, but perhaps not.

Key takeaways

Making an additional plant-based option available did not meaningfully change meal selections. This held for both a long-standing PMA option (Sofritas) and a novel one (Chickā€™nitas). Reducing meat consumption is a hard problem, and achieving price-, taste-, and convenience- parity wonā€™t obviously do the trick either. 

Technology will likely play a large role if we are to transition away from factory farming, as it has for past externalities; but we also see an important role for culture and attitudinal change, as well as changing the choice architecture of meal environments and actively promoting plant-based options.

LewisBollard @ 2025-09-08T16:19 (+9)

Thanks for doing and publishing this study! It's so helpful to get a clearer picture on this, even if we don't like the answers. As a validation of your findings, in 2015, Chipotle told Vox that sofritas were 3.5% of sales, very close to your 3.8%. 

Seth Ariel Green šŸ”ø @ 2025-09-08T17:12 (+13)

Interesting! I believe I missed that interview, although a rep told the Times that same year that "sofritas accounts for about 3 percent of fillings." 

I recently learned that Steve Ellis (Chipotle founder) tried predominantly plant-based fast casual in 2024; apparently it didn't work out (although I'm still seeing a Yelp page?) and this winter he told Eater that  ā€œveganism...is very polarizing, Iā€™ve learned.ā€

 In a separate interview Ellis said ā€œI think people will eat more plant-based diets and make that part of their life if there are better options,ā€ [emphasis added], which I agree with. But I basically think that that no single effort is likely to change the game, in part because the most effective interventions are unusually challenging to scale.

(Also, Chipotle tried out free shipping on plant-based options in 2021, which is an intervention I'd love to formally evaluate! But I'd bet that appetite for that kind of thing is lower nowadays. The whole market is a long way from where it was then.)

Dorsal Heart @ 2025-09-10T00:54 (+3)

I appreciate y'all studying this and helping us learn more about what we can do to advance plant-based options, and I especially love that it is open access. I do have some questions tho. 

(A) "Despite widespread optimism, simply increasing the number of PMAs on restaurant menus may not consistently reduce meat selection." Is this a reasonable expectation of any new food ingredient? Are there studies that show just adding a new ingredient, any ingredient whether plant-based or not, to a menu would result in uptake of it? 

It seems to me, and I could be missing something, that this study might be studying the habitual nature of consumers, and less so their preferences about plant-based meat. A possible way to have controlled for this would be to have a new animal meat item and see how many people chose it. 

If your general model of consumers, or people generally, is that they are cognitive misers, it seems that model would predict this result regardless if the additional items were plant based or meat based. So the casual factor would be habit, not the kind of protein. That would still be informative for plant-adoption and be a large barrier, but it be saying less about plant-based preferences and more about just needing to do proper UX and marketing. 

(B) I saw y'all did awareness checks on what people thought the point of the study was, but curious why y'all didn't do an awareness check on whether they noticed if there was a plant-based option in the menu or not? Like how many people are taking the time to read the small grey on white text? And with steak or chicken, you don't need to because those are familiar and instantly understood. Without that, it is unclear to me, whether you are testing people's preference for plant-based options or their ability to notice new items. 

Seth Ariel Green šŸ”ø @ 2025-09-10T12:17 (+2)

Hi Dorsal, good questions:

  1. In general, as an economist friend put it, "Changing options is a very strong intervention, like mechanically there should be an effect." So I would expect a new meat option -- BBQ chicken or whatever -- to attract customers. But you are right, we don't know that. On the other hand, our question was whether adding a chicken analogue would attract customers away from meat-based options, so whether a meat option would have also attracted customers is not really apropos of our estimand. It might help put our results in context, but it's not the theoretical quantity we're after. And there's a lot to be said for keeping a study focused. Another thing manipulated means either a smaller sample per treatment arm or a more expensive experiment. Always we are triaging.
  2. That would have been a fine thing to check, but in the online ordering context we were trying to simulate, you also view the options without necessarily "taking the time to read the small grey on white text," so if they miss the new option, that's arguably an element of experimental realism. Also, as Lewis points out, we have reason to think our numbers are broadly in line with what people are actually ordering, which is some evidence that people were actually reading. However it might be interesting to do a follow-up where someone actively promotes the new PMA, which restaurants sometimes do. That would be a fine paper, but also a different experiment aimed at a different estimand.