photo_camera Jo-Anne McArthur / We Animals
Resident rescued pigs at Farm Sanctuary
Survey
Animals, Food, and Technology (AFT) Survey: 2021 Update
Ali Ladak and Jacy Reese Anthis  •  July 15, 2022

For full details on the methodology and previous years’ results, see our 2020 and 2017 reports. Links to survey and code for this report can be found here. The dataset can be downloaded here.

Please cite this report as follows: Ladak, Ali; Anthis, Jacy Reese (2022). Animals, Food, and Technology (AFT) Survey: 2021 Update. PsyArXiv. https://doi.org/10.31234/osf.io/3v6ku

Summary

The Animals, Food, and Technology (AFT) survey tracks attitudes towards animal farming and animal product alternatives in the US. As in previous iterations, we found strong opposition to various aspects of the animal farming industry, with a majority of people reporting discomfort with the industry (74.6%), and relatively strong support for a range of seemingly radical policy changes, such as banning slaughterhouses (49.1%), factory farming (52.7%), and animal farming as a whole (38.8%). Regional analysis indicated that the West North Central region (the nation’s “Agricultural Heartland”) has the least Animal Farming Opposition (AFO) and Animal Product Alternatives Support (APAS). While AFO in the US as a whole fell in 2020, it recovered close to its 2019 level in 2021, though the overall trend and the difference in AFO between 2017 and 2021 is not statistically significant.[1] APAS has gradually fallen between 2017 and 2021, though this trend is non-significant.

Table of Contents

Introduction

2017–2021 Trends

2021 Results

Means and Distributions

Regional Analysis

Introduction

In October 2021, we conducted the fourth iteration of the Animals, Food, and Technology (AFT) survey on attitudes towards animal farming and animal-free food with 1,532 US adults, census-balanced to be representative of age, gender, region, ethnicity, and income.[2] Here we report:

  1. The trends in responses between 2017 and 2021
  2. The means and distributions from the 2021 survey
  3. A regional breakdown of the key findings from the 2021 survey

Our methodology remained largely the same as in previous years and can be found in our 2017 and 2020 reports. Any deviations from this are included in the text. Links to survey analysis code for this report can be found here. The dataset can be downloaded here.

2017–2021 Trends

Figure 1 presents trends in the mean responses and confidence intervals for each survey question for the four years we have collected data, weighted to account for differences between our sample and the US population on age, gender, region, ethnicity, and income using the most recent American Community Survey census data. The chart also shows the trends in the two scales we developed from these questions: Animal Farming Opposition (AFO) and Animal Product Alternatives Support (APAS). Definitions of these and text of all questions can be found here.

Figure 1: Animals, Food, and Technology survey 2017–2021 trends. Please click on items on the legend to show trends for questions of interest. This figure is best viewed on a laptop or desktop screen.

The plots suggest that AFO has fluctuated over the period. After falling between 2019 and 2020, it recovered close to its 2019 level in 2021. However, regression analysis of AFO over time suggests that the overall trend in AFO between 2017 and 2021 was non-significant (b = 0.023, SE = 0.023, p = 0.313), and the difference between 2017 and 2021 is also non-significant (b = 0.13, SE = 0.07, p = 0.073). APAS gradually fell over the period, though this trend was also non-significant (b = -0.03, SE = 0.02, p = 0.162). The full results of the time trend analysis can be found here. We report the key trends below.

2021 Results

Means and Distributions

Figure 2 presents the weighted means and distributions of responses for the questions in the 2021 survey.

Figure 2: Animals, Food, and Technology survey 2021 weighted means and distributions. Please use the dropdown menu to show results for questions of interest. The intervals for AFO score and APAS score are open on the left and closed on the right. This means, for example, that an AFO score of 2 will be included in the interval 1–2 and not 2–3. This figure is best viewed on a laptop or desktop screen.

We report key findings for respondents who expressed an opinion below.

Regional Analysis

Figure 3 presents weighted means and confidence intervals for key variables, broken down by region. We present the data for the nine US census divisions, which strikes a balance between the level of detail and sample size in the regions.[3] In the future we may combine regional data from the different survey waves to provide a larger sample and a more detailed breakdown (for example, by state).

Figure 3: Animals, Food, and Technology survey 2021 regional breakdown. Please hover over the map to view regional data. This figure is best viewed on a laptop or desktop screen.

We also conducted weighted t-tests to statistically test for differences across the regions. We report the key findings below. Note that these results describe raw differences between the regions and do not control for other factors, such as politics, that may contribute to the regional differences. Some of these results also do not hold after correcting for multiple comparisons.[4] The full results, including results of regressions for AFO and APAS controlling for a range of demographic variables, can be found here.


[1] Here and throughout this report we use the conventional threshold of p = 0.05 to distinguish statistically significant vs. non-significant results; some of these results do not hold after correcting for multiple comparisons.

[2] The sample is larger than in previous years because of a change in how the sample was collected over time (i.e., not collecting hardest-to-recruit demographics first) and an oversampling by one of the data collectors. These changes should not substantially affect results aside from higher statistical power: The design effect was 1.33 and the effective sample size was 1152.

[3] The nine divisions are split as follows:

[4] To correct for multiple comparisons we held the false discovery rate at 10% for the t-tests for each variable, following Benjamini and Hochberg (1995).


Subscribe to our newsletter to receive updates on our research and activities. We average one to two emails per year.