Increasing Donations Through Appeal Types, Exposure, And Donor Characteristics

Companion animal charities tend to receive the vast majority of the donations made to animal welfare groups, dwarfing donations to farmed animal causes. Although it is easy to assume that people just care more about companion animals, our recent study of donor characteristics found that there may be more to it than that: differences in messaging or even a simple lack of exposure to farmed animal charity appeals may partially account for the differences in donations. When messaging and exposure were kept the same for companion and farmed animal charities, the proportion of people who gave to companion animals remained high, but the proportion who gave to farmed animals more than doubled relative to their past behavior.

In the current study, we considered whether the difference in donations to companion and farmed animals is due in part to messaging style and amount of exposure. We also looked at the role of previous donations to animal charities.

Key Findings

• Farmed animals received almost as much money as companion animals when donors were given the opportunity to support them: 86% of the amount. This suggests that farmed animal charities may benefit from just getting their message out more. However, it is worth noting that the farmed animal appeals in this study were “diet-neutral”: There was no mention of diet, veganism, or the donors’ behavior. We recommend that donation appeals to a general audience (vs. just vegans) focus on the reduction of animal suffering and not refer to human diet or lifestyle, because it sidesteps the “meat paradox”: the fact that many people care about animals but still eat them. It would likely be hard for those individuals to provide financial support to a cause that explicitly works against their own behavior.

• People who have previously donated to any animal charity gave 92% more money to farmed and companion animals than people who had not. Therefore, the more it costs a charity to access each potential donor, the more cost-effective it becomes to tap into this group. For farmed animal charities in particular, previous donors to any animal cause are a valuable target group, as most will likely be new to farmed animal support.

• There was no difference in donations whether the appeal used an identifiable victim or statistical victims. This finding is similar to previous research with wild animals (Thomas-Walters & Raihani, 2016). Surprisingly, though, even appeals with no descriptive text at all performed as well as the rest. This finding points to minimal appeals as the most cost-effective messaging strategy for donations, as it suggests that lengthier ads may be unnecessary.

The Study In Brief

This is a short overview of the study. Full procedural details and the survey instrument are available on our project page on the Open Science Framework.

Companion animal charities often present rescued cats and dogs as individuals with names and stories, whereas farmed animal charities often focus on the number of animals affected. 

To investigate the impact of messaging style, we looked at four different appeal types:

1. An identifiable animal appeal (e.g., “This is Susie”), 
2. A statistical appeal (e.g., “Over 10 billion farmed animals”), 
3. A combined appeal (e.g., “This is Susie. Like over 10 billion other farmed animals…”), and 
4. An appeal with no additional text at all. 

Each of the appeal types was presented for both farmed animals and companion animals, so there were eight conditions in total. This design allowed us to directly compare the different types of appeal and to see whether they were more or less effective for farmed or companion animals. 

Towards the end of the study, participants were given a 50-cent bonus and told that they could donate as much or as little of it as they liked to the Animal Welfare Institute. This donation was our outcome measure.

The Animal Welfare Institute was the charity used in all appeals in this study because they support both farmed and companion animals, and have a simple, descriptive name. They agreed that their name and logo could be used in this way, and we passed along all participant donations to them. 

Figure 1: An example of one of the appeals: The combined appeal for farmed animals

While we didn’t explicitly test whether a lack of exposure to farmed animal charities could account for their historically low number of donors, we held exposure constant. Therefore, this study provides some evidence for the role of exposure because we can compare the amount donated to farmed animals in this study to what they typically get (i.e., 0.8% of all donations to animal charities; Charity Navigator, as cited in Animal Charity Evaluators). 

We also asked participants about their donation history and compared the donations of people who had previously donated to animal causes with those who hadn’t. 

As a note, our sample, which we recruited through Positly, was slightly different from the general population in a few ways. For example, people of color were under-represented, and people aged 25 to 44 were over-represented. Because of these differences, the means shown in the graphs below are weighted to be more representative. However, we conducted the statistical tests on differences between the groups on the unweighted data because the differences from population values were not huge (all less than 10%) and weighting introduces another kind of bias in inferential testing.

Results

In this section we describe the results in broad terms, explaining which differences are significant and which are not. The error bars on the graphs also provide a visual indication of significance. Because these are statistical estimates, there is a degree of uncertainty about the true value—the bars show the 95% confidence interval for each estimate. The more the confidence intervals for two estimates overlap, the less likely it is that they are significantly different. 

For full statistical details, see the “More Details” section at the end of this report.

Are There Differences Between Statistical and Individual Appeals?

The question of whether statistical or individual appeals are more effective is important because the evidence from past research on human and animal appeals is mixed (e.g., Hart, Lane, & Chinn, 2018; Mercy for Animals, 2016; Västfjäll, Slovic, Mayorga, & Peters, 2014). 

So what did our results show? There were no significant differences among the four types of appeal for either farmed or companion animals. That is, when we compared the four farmed animal appeals to one another and the four companion animal appeals to one another, there were no significant differences between any of them.

This set of findings means, for example, that an appeal that had both an identifiable victim and relevant statistics did no better than an appeal that simply asked for a donation with no descriptive text at all! 

At this point, it is important to note that our study was well-powered, meaning that we had enough participants to detect reasonably small differences if they exist (more details below). In addition, this lack of difference replicates the findings of the most relevant previous study (Thomas-Walters & Raihani, 2016)—a controlled experiment looking at donations to wild animals. Those authors, like us, found that using identifiable victims was no more effective than using statistical victims.

Do Farmed Animals Still Receive Less Money If They Have The Same Exposure As Companion Animals?

Yes, but the difference was much smaller than we expected! 

Past work by Charity Navigator has shown that 66% of donations to animal charities go to companion animals and just 0.8% go to farmed animal charities (see Animal Charity Evaluators, 2016). In our recent donor study, we found that the percentage of people who donate is similarly skewed: 82% of donors had given money to companion animals in the past year, versus just 12% to farmed animals. 

In the current study, participants gave an average of 16.4 cents (weighted) to companion animals. If we were to guess from that how much farmed animals should get based on Charity Navigator’s statistics, we would expect them to receive an average donation of just 0.2 cents—less than half a cent.

Contrary to that prediction, farmed animals received an average donation of 14.1 cents! That’s great news for farmed animals: when put in front of potential donors on an equal footing, farmed animals appeals appear to do almost as well as companion animal appeals.

Were There Differences Between Those Who Had Donated To Animal Charities Before And Those Who Had Not?

Yes, there was a large difference, as you can see in the figure below. People who had previously donated to animal charities gave almost twice as much on average as people who had not!

We didn’t ask participants which types of animal charity they had previously donated to, because we didn’t want them to be overly influenced by their previous behavior. But everything we know about donations to animal charities suggests that most of those previous donations likely went to companion animal charities, so this finding is particularly important given that past donors gave substantially to farmed animals. 

Although past donors gave less to farmed animals than to companion animals, as shown below, their donations were still much larger than those from people who had not previously given to animals. In short, previous donors to any animal cause are likely potential farmed animal donors.

Were The Findings The Same For Number Of Donors?

Apart from the dollar value of the donations received, the number of donors is an important statistic for charities. Each new donor represents an opportunity for future donations as well as other forms of animal advocacy.

For that reason, we repeated each of the above analyses using the proportion of people who donated as the outcome variable of interest. The pattern of results was the same as for donation amount: The percentage of donors did not differ significantly between the different types of appeal, more people donated to companion animals than farmed animals, and previous donors were more likely to donate.

Implications

Similar to previous research with wild animals (Thomas-Walters & Raihani, 2016), our study revealed no difference in donations between using an identifiable victim or statistical victims for farmed or companion animals. Surprisingly, even appeals with no descriptive text at all performed as well as the rest. This finding points to minimal appeals as the most cost-effective messaging strategy for donations, as it suggests that lengthier ads may be unnecessary. 

This study also found that farmed animals received almost as much money as companion animals when donors were given the opportunity to support them. The current study gives us hope that outreach for farmed animals can be successful, thus bringing important capital into the movement: We think that farmed animal charities would probably benefit from getting their message out there more!

Finally, we found that people who have previously donated to animal charities gave substantially more money to both farmed and companion animals than people who had not. This group donated 92% more on average than people who had not previously donated. Therefore, the more it costs a charity to access each potential donor, the more cost-effective it becomes to tap into this group. For farmed animal charities in particular, we recommend targeting previous donors to any animal cause when possible—they will likely be new farmed animal donors.

That said, there are far more non-donors to animal causes than donors, and bringing in new supporters to the movement is important. Almost half of the non-donors gave to farmed animals, with an average donation of 11.5 cents. Therefore, targeting people who have never previously donated to animal causes is a viable strategy, and will lead to the long-term growth of the movement.

It is also worth noting that the farmed animal appeals in this study were “diet-neutral”: We just asked for a donation without any mention of diet, veganism, or the donors’ behavior. This approach may allow advocates to get around the “meat paradox,” where people know that eating meat causes harm to animals, and have mental defenses that they use to justify it (Piazza et al., 2015). 

For example, it may be tempting for vegan organizations to frame donation appeals in terms of the big goals, like a plant-based future or an end to using animals for food. However, we suggest that those types of messages may reduce donations from non-vegans, because this group would have to face the paradox of donating to end something that they participate in. Although forcing people to face the paradoxes in their behavior and attitudes can be good in a behavior change campaign, it may be counter-productive in a donation request. This is because the easiest way for an omnivore to resolve the feelings of dissonance that arise from being asking to donate to a plant-based future is to choose not to donate: if they do give money, they have to then wrestle with their behavioral choices as well. That said, this idea also needs to be tested in future studies specifically designed to directly compare messages that reference diet to those that don’t.

All in all, we recommend that donation appeals to a general audience (vs. just vegans) should focus on the reduction of animal suffering and not refer to human diet or lifestyle. Because people often infer their attitudes from their behaviors (Bem, 1972), we think that if we get people to donate to animal causes, it may then help lead them to make additional animal-positive changes in their lives as well.

How Confident Can We Be In These Results?

Were The Donation Amounts Too Small To Matter To People?

While the donation amounts were indeed small—at most 50 cents— they need to be considered in the context of the study. People completed the survey itself in exchange for just 50 cents. So to argue that the donations were too small to be meaningful to participants would imply that the incentive to participate wasn’t meaningful either. Yet we found over 2,600 people willing to complete the survey within a few hours. Had they chosen to keep the 50-cent bonus, they could have doubled their wage, yet many did not. All that said, we recognize that the dynamics in the study may have been different if larger donations were involved and we encourage additional research in this area.

What If The Farmed Animal Appeals Were Extraordinary, Leading To Higher Donations Than Usual?

We doubt this was the case—the pictures we used were similar to real appeals in the animal welfare space. We also chose similar companion and farmed animal images. Future Faunalytics studies will look more carefully at imagery, examining how it affects donation amounts. 

What If People Gave More Because We Gave Them A Bonus?

The money participants could donate was given to them as a bonus. This raises the question of whether they may have donated more than usual because it felt like “free money” or because of the norm of reciprocity: when a person is given something, they may feel the need to return the favor in some way (Gouldner, 1960). In our study, we both gave people a bonus and asked them if they wanted to make donation, meaning the norm of reciprocity may have increased the amount people gave. However, unless we think this norm would operate differently in companion than farmed animal appeals, or in statistical victim appeals than identifiable victim appeals for example, it would not make a meaningful difference to those results. However, we also considered donation results in the current study relative to past research—this comparison would be more likely to be affected by the norm of reciprocity, so these results should be replicated using additional donation methods. 

A Final Word Of Caution

We used an experimental design and collected a large sample in order to test our research questions, and we had sufficient power to be able to detect a small-to-medium effect. However, no individual study is perfect or definitive. 

All told, if you have reason to believe that your audience prefers one type of appeal over another, it might be worth checking to see if our findings hold true in your context. We recommend A/B testing your ads or email subject lines to see if your hunches are correct. We are also available through our office hours if you would like to discuss something along these lines.

More Details

Participants And Power

All participants were recruited through Positly, a company that screens out fraudulent and low-quality participants. 

In addition, we included our own internal data checks in keeping with the Faunalytics Data Quality Assurance Plan. We removed relatively few participants from the sample due to data quality checks (n = 20 removed), and have confidence that those included in the analyses are valid participants who were attentive to the study tasks.

We had a total of 2,630 participants after data cleaning, determined by our pre-registered power analysis. This is enough power to detect a small-to-medium effect size of d = 0.3 when comparing between appeal types. For additional details on the measures, power analysis, analysis plan and more, please see the pre-registration documents on the Open Science Framework.

Analyses

The following are the statistical results for each of the findings reported in the sections above, labeled by section. Given the large number of pairwise comparisons in this study, we corrected for the false discovery rate (FDR) using the Benjamini-Hochberg procedure, with the FDR set at 5%.

• Are there differences between statistical and individual appeals? 
  • Within the set of farmed animal appeals, there were no significant differences in mean donations among the four types of appeal, all ps > .76. (This was still true even without the FDR correction.)
  • Similarly, within the set of companion animal appeals, there were no significant differences in mean donations among the four types of appeal, all ps > .79. (This was still true even without the FDR correction.)
• Do farmed animals still receive less money if they have the same exposure as companion animals?
  • Yes, participants gave more money on average to companion animals than farmed animals (M_CA = 15.9 vs. M_FA = 13.8, unweighted), t(2623) = 2.80, p = .005. 
• Were there differences between those who had donated to animal charities before and those who had not?
  • Yes, past donors gave more money on average than past non-donors (M_PD = 22.9 vs. M_PND = 12.0, unweighted), t(1065) = 12.74, p < .001.
  • This was true for donations to farmed animals (M_PD = 20.9 vs. M_NPD = 11.3, unweighted), t(547) = 8.11, p < .001.
  • It was also true for donations to companion animals (M_PD = 25.0 vs. M_NPD = 12.8, unweighted), t(520) = 10.04, p < .001.
• Were the findings the same for number of donors?
  • Omnibus chi-squared tests indicated no effect of appeal type on the number of donors for either farmed animals, Χ² (3, N = 1298) = 1.4, p = .71, or companion animals, Χ² (3, N = 1332) = 2.9, p = .40.
  • Marginally more participants donated to companion animals compared to farmed animals, Χ² (1, N = 2630) = 3.8, p = .05.
  • Past donors also gave more frequently than non-donors Χ² (1, N = 2630) = 128.9, p < .001.
  • This was true for farmed animals Χ² (1, N = 1316) = 57.0, p < .001.
  • It was also true for companion animals Χ² (1, N = 1314) = 72.4, p < .001.