Pork Or Pig? Beef Or Cow? Implications For Advocacy And Research
Many animal advocates have discussed the implications of referring to animals and animal flesh by common euphemisms like beef, pork, and livestock, rather than explicitly as cow meat, pig meat, and animals. For example, Joan Dunayer and Melanie Joy have written extensively on the topic of speciesist language and, as Joy refers to it, “linguistic deception” via euphemism.
This perspective is supported by research. Euphemisms allow people to distance themselves from thoughts of where their food comes from, and unpleasant feelings associated with that. And research has shown that using the terms “cow” and “pig” on a menu instead of “beef” and “pork” increased empathy and disgust and reduced willingness to eat meat. They found similar results for describing cows as being “slaughtered” or “killed” versus the euphemism “harvested.” People who read the euphemism felt less empathy for the cows.
We at Faunalytics wondered how far this effect would go. In part to see if just referring to animal meat explicitly could have an immediate effect on attitudes (which could be good), and in part because we thought the choice of term might affect study results (which could be bad if different studies use different terms). We ran two randomized controlled trials (RCTs) to look at this question.
The Studies In Brief
More details at the bottom of the post.
In both RCTs, participants were randomly assigned to one of two wording conditions: standard (euphemistic) wording or alternative (explicit) wording. The two studies used different question types, but in both, participants assigned to standard wording were asked about their consumption of “pork” and “beef” while participants assigned to explicit wording were asked about “pig” and “cow.”
In the first study, after the diet question with one of the two wordings, participants were then asked to indicate the extent of their agreement or disagreement with the statement, “It is important to me that animals used for food are well cared for.”
In the second study, the follow-up agreement question used the statement: “Farm animals deserve the same consideration as pets and other animals, even if it means that meat costs more.”
In both cases, we hypothesized that participants in the explicit wording condition (vs. the standard wording condition) would report eating less beef/cow and pork/pig and express more agreement with the animal welfare statement.
To our surprise, those hypotheses were not supported in either study. None of our pre-registered analyses showed significant differences between the euphemistic and explicit wording. In other words:
- In Study 1, the percentage of people who said they had eaten beef or pork in the past year was no different from the percentage who said they had eaten cow or pig;
- In Study 2, participants’ reported frequency of consumption did not differ based on whether they were asked about their frequency of consuming beef and pork or cow meat and pig meat;
- In Study 1, the average response to the statement “It is important to me that animals used for food are well cared for” did not differ based on whether people had seen the standard euphemisms or the explicit terms;
- In Study 2, the average response to the statement “Farm animals deserve the same consideration as pets and other animals, even if it means that meat costs more” also did not differ.
Based on these findings, we would caution that word choice in many typical advocacy contexts may be more symbolically important than strongly influential. Although previous research showed that using explicit language at the moment of food choice can influence that choice, these studies suggest that it may not have much influence on attitudes outside of that context.
So an advocate’s choice of pig versus pork may not add much oomph to their argument. Even still, this doesn’t mean that you shouldn’t use accurate terminology that provides a subtle reminder of where meat comes from—please do! It is possible that repeated use might have a noticeable effect. These studies don’t address that.
It would be ideal if referring to animal flesh by its explicit name even once would make people feel immediately more pro-animal—and to be honest, we thought it would, even if in only a small way. Unfortunately, there is no evidence that that is the case.
As with any intervention, chances are that the more strongly the meat-animal connection is made and the more it matters, the stronger its effect on attitudes and behavior (at least to a point—be mindful of reactance). These RCTs were conducted in the context of survey wording, which is important because of its prevalence in advocacy research. To researchers, it should be reassuring that differences in terminology on surveys are unlikely to introduce bias.
How Confident Can We Be In These Results?
Number of Participants
We powered these studies to be able to detect a small difference (d = 0.2; w = 0.1; f2 = 0.02) at the standard 80% level. In lay terms, that means that there’s a 20% chance that each study individually could be wrong in saying that there’s no difference between standard euphemisms and explicit language, just due to random chance. However, the probability of both studies being wrong is only 4%, so we can be quite confident that random chance isn’t to blame.
We really thought that there would be a difference between the standard euphemisms and explicit terms, so when the first study didn’t show significant differences, we thought it might be something about the specific questions we used. That’s why we ran the second study.
In particular, we thought that the agreement statement from Study 1 may have been affected by a ceiling effect (most people tending to agree whether they’ve just read explicit terms for animal meat or not). We also thought that the categorical consumption measure (have eaten or not over the past year) might be too straightforward to be impacted by any discomfort caused by the explicit terms.
Because of those concerns, Study 2 used a more nuanced agreement statement that wouldn’t pull for agreement as strongly and measured responses to it on a 7-point scale rather than a 5-point scale. It also used a measure of consumption frequency rather than a checklist. As described above, even with those changes, there was still no difference between conditions.
These studies were conducted on Mechanical Turk, which may concern some readers given recent problems with bots or study farming. However, these studies were conducted in March and May of 2018, well before researchers began reporting an increase in problems on MTurk. In addition, we excluded data from accounts with duplicate IP addresses or IP addresses from outside the U.S.
For details of the measures, power analyses, analysis plans, and more, please see the pre-registration documents for both studies on the Open Science Framework. In addition, the Study 2 document contains more details about our concerns with the specific questions used in Study 1, as discussed above.
- Final sample: n = 775 Mechanical Turk participants with approval rating greater than 90% (after excluding 8 IP duplicates and 20 located outside the U.S.)
- Of the participants in the standard wording condition, 89.9% said they have eaten beef in the past year. This proportion did not significantly differ from the 87.7% of participants in the explicit wording condition who said they have eaten cow in the past year, χ2 (775) = 0.91, p = .34.
- Of the participants in the standard wording condition, 85.1% said they have eaten pork in the past year. This proportion did not significantly differ from the 80.2% of participants in the explicit wording condition who said they have eaten pig in the past year, χ2 (775) = 3.22, p = .07.
- Mean responses to the 5-point attitude item (“It is important to me that animals used for food are well cared for”) were quite high, above the agree scale anchor. There was no difference between the standard wording (M = 4.09, SD = 0.89) and explicit wording (M = 4.10, SD = 0.96) conditions, t(773) = -0.17, p = .87.
- The differences observed in this study were in the expected direction and close to significant for pig/pork. This inspired us to run Study 2, but because those results didn’t replicate, those small differences shouldn’t be interpreted as meaningful.
- Final sample: n = 799 Mechanical Turk participants with approval rating greater than 90% (after excluding 13 IP duplicates and 22 located outside the U.S.)
- There was no significant difference in participants’ reported consumption of beef (standard wording) versus cow meat (explicit wording), as determined by ordinal regression, Wald χ2 = 0.08, p = .78. The proportions by condition were:
- There was no significant difference in participants’ reported consumption of pork (standard wording) versus pig meat (explicit wording), as determined by ordinal regression, Wald χ2 = 0.03, p = .86. The proportions by condition were:
- Mean responses to the 7-point attitude item (“Farm animals deserve the same consideration as pets and other animals, even if it means that meat costs more”) did not significantly differ between the standard wording (M = 4.60, SD = 1.62) and explicit wording (M = 4.67, SD = 1.65) conditions, t(797) = -0.56, p = .58.