Does Animal Advocacy Messaging Influence Support for Policies Affecting Wild-animal Suffering?
This is a technical document detailing the results of a study we did on Mechanical Turk in 2016. You can see a brief, accessible summary of the results in our related blog post.
We carried out a survey to identify possible effects of animal advocacy messaging on people’s support for policies that would affect the suffering of wild-animals. Six hundred twelve (612) participants were randomly shown one of three leaflets: (1) one advocating dietary change on the basis of cruelty to farmed animals, (2) one advocating dietary change on the basis of environmental damage, and (3) a control leaflet advocating for volunteering and donating to homeless shelters. They were then asked how much they supported or opposed seven different policy decisions, and their responses were scored according to how we expect the decisions might affect the suffering of wild animals.
For questions 1-4, an answer of “Strongly Support” earned 3 points, while an answer of “Strongly Oppose” earned -3 points, with intermediate answers being interpolated evenly. This was reversed for questions 5–7. See the scoring table below:
|WAS Points (+/-)
For questions 3, 4, 6 and 7, our scoring depends on the assumption that most lives in the wild are not worth living, which seems to be a common view among those who have researched this issue, and is the same view most animal advocates have of animals in factory farms. If you want to avoid this assumption, you could solely consider responses to questions 1, 2, and 5. See the methodology pre-registration for more details on how the study was designed.
We committed to comparing the mean scores of each treatment group using t-test for statistical significance (α=0.05), excluding the scores for questions 6 and 7 if they correlated below R2=+/-0.3 with questions 1–5. They correlated well below this (Table 1), and so were excluded. All other analyses reported here should be treated as exploratory.
Results and Comments
Differences in groups on WAS scale
Summing scores across questions 1–5, we found no difference between the Control and Cruelty groups (p=0.996) (Figure 1). The Environment treatment performed worse (had scores which indicated less concern for WAS), and this was nearly significant (p=0.060). Exposure to the Environment message shifted the average participant down 0.7 points, the equivalent of changing an answer from Somewhat Support to Undecided (or Strongly Support to Support) on one of five questions, 70% of the time.
Figure 1: Group means by question. Error bars show the standard error of the mean.
Scores were weakly correlated between questions (Table 1). The only inter-question correlation >0.3 was between questions 3 and 4 (r=0.503), which were both examples of habitat destruction. Questions 6 and 7 did not correlate with the others exceptionally weakly, but were nevertheless excluded from the main analysis according to our criteria.
Table 1: Correlations (r) between questions.
Suppose you live near a forest. This forest has a large deer population due to a lack of predators, which leads to starvation and illness in many deer. Your city council is considering a proposal to distribute birth control to these deer to painlessly limit their population. Would you support or oppose this proposal?
The Control group was most supportive of distributing contraceptives to wild-animals starving due to overpopulation, but not significantly more supportive than the Cruelty group (p=0.459). Participants shown the Environment message were least supportive (Control vs. Environment: p=0.073). This was in line with our expectations that the Cruelty group would be most supportive, though not by much, and the Environment group would be most averse to interfering in the natural lives of these animals.
Suppose rangers on a wildlife preserve come across an elk that has been injured by a wolf attack. The deer could survive with medical attention. Would you support or oppose a decision by the rangers to help the animal?
The Environment group was less supportive than Cruelty (p=0.040) or Control (p=0.055) of giving veterinary attention to injured wild animals. The Control and Cruelty groups were about equally supportive (p=0.874). As with question 1, it is sad but unsurprising that environmental preservation messaging seems to make people less supportive of helping wild animals.
Questions 3 and 4
A developer has proposed building a shopping mall in your area. The mall would generate $5 million per year in tax revenues for the local government, but its construction would require paving over a region that is currently wetland. Would you support or oppose the mall proposal?
Much of the world’s palm oil is grown in Malaysia and Indonesia, where it provides income and employment to local people. However, its cultivation sometimes requires burning down rainforests. Would you support or oppose investments in palm oil?
There was no significant difference between the control and each treatment group in support for paving over wetlands to build a shopping mall (Control x Cruelty: p=0.335; Control x Environment: p=0.423) or burning rainforest to grow palm oil (Control x Cruelty: p=0.823; Control x Environment: p=0.429).
It is surprising that the Environmental message had such a minor effect here. Perhaps it is because the public is already saturated with ‘green’ messages, so habitat destruction falls too far outside the Overton Window for our brief appeal to have much influence. Indeed, these questions do appear to be especially polarising; while their mean scores hover around “Somewhat Oppose,” they both have the highest number and ratio of “Strongly Oppose” vs. “Strongly Support” answers, with the close exception of Question 6.
At the same time, the connection between habitat destruction and wild animal welfare is not immediately obvious. Since habitat destruction might inflict short-term suffering, support could also vary depending on how one discounts present-versus-future or anthropogenic-versus-natural wild animal suffering. Of course, the average survey participant probably isn’t giving the dilemma this much thought, and could just think of the present/anthropogenic damage without considering other outcomes.
Rabbits are an invasive species in Australia, and their eating behavior has reduced native plant populations. The government is considering poisoning some of the rabbits to reduce their population and preserve the native plant-life. Would you support or oppose this program?
The Cruelty group was least supportive of culling invasive rabbits to preserve native plants (Control x Cruelty: p=0.199; Cruelty x Environment: p=0.077), while the Environment group was most supportive. This question traded off concern for animals against environmentalism, but unlike in Questions 3 and 4, the wild-animal-friendly answer—opposition to the culling program—does not entail any direct harm to animals. Our scoring on this question does not assume that most lives in the wild are net-negative.
Questions 6 and 7
Terraforming is a hypothetical process in which a planet like A2:L614 would be modified in order to make it habitable by plants and eventually animals, including humans. Assuming it could be done cheaply, would you support or oppose terraforming Mars?
In the distant future, an advanced civilization is interested in studying the evolution of life on Earth. It intends to do so by running a large computer simulation of Earth’s history, including the creation of ‘digital organisms’ that are functionally identical to biological organisms. Would you support or oppose this undertaking?
On far-future questions, both Treatment groups scored significantly higher than the Control (Control x Treatment: p=0.009), but very similarly to each other (Cruelty x Environment: p=0.885). It is not surprising that environmentalist messages would make people less supportive of terraforming, preferring to leave Mars in its natural state. The creation of digital organisms and its connection to environmentalism or animal welfare is not necessarily intuitive, so it is interesting to see a consistent and significant effect of animal advocacy messaging.
As to how we weigh these responses in our decisions on which messages to spread, we should keep in mind that these far-future questions are vastly greater in scale than any of the other questions, but must also be discounted because of the uncertainty over whether humans will ever have the opportunity to make such decisions, and the amount of time and social change between today’s animal advocacy and the people who will make them.
This study provides some evidence that encouraging people to reduce their consumption of animal products on environmental grounds makes them less supportive of intervening in nature to alleviate wild-animal suffering. Our exploratory analyses suggest that this is true with and without the assumption that most wild animals live net-negative lives. However, the fact that the average respondent scored positively on all questions that did not make this assumption (1,2,5), but negatively on all questions than did (3,4,6,7), suggests that they as whole may be more wild-animal-friendly than appears, while believing that wild animals live lives worth living in their natural habitats.
These results suggest that animal advocates who are concerned about wild animals should place less emphasis on the environmental harms of animal agriculture. Our findings should allay concerns that cruelty-focused farmed animal advocacy messages may reduce people’s willingness to help wild animals. Of course, these results should be taken as just one piece of evidence in the overall decision of how to best advocate for animals.
You can download the raw data from this study here.
Edit 10/4/2016: The r table in this post previously used Spearman’s R and is now corrected. The attached spreadsheet had some incorrect labeling (whether questions were “Support = positive” or “Support = negative”) and is now corrected.