Which Farmed Fish Welfare Issues Does the Public Care Most About? (Neutral Messaging)

The animal protection movement is in great need of research into effective fish advocacy tactics for two main reasons:

There are many issues that impact the well-being of farmed fish. In this study we wanted to see which of these issues generated the most public support across several categories of response.

We tested six treatment texts against a control (a blank screen without any text). Each treatment was a single paragraph that described one issue impacting the welfare of farmed fish. See the supporting documents section at the bottom of this blog for a link to the full treatment text. The six issues we focused on were as follows:

Treatment 1: Crowding

Treatment 2: Slaughter

Treatment 3: Handling

Treatment 4: Feeding Practices

Treatment 5: Poor Breeding

Treatment 6: Disease and Water Quality

Importantly, the treatment texts were factual statements about welfare issues and did not discuss any specific qualities of fish sentience or highlight individual fish. We presented the texts in the way a scientist or journalist might present the facts, as opposed to the way an animal advocacy group would present the facts. To have done otherwise could have had a meaningful impact on the responses generated. In a subsequent study, we will look at similar questions but with text blocks presented in a style more typical of animal advocacy groups.

We recruited 2,611 participants from Amazon Mechanical Turk to complete the survey and used 2,420 complete responses. There were around 1,200 responses for the control group and 200 responses for each treatment group.

After the participants read the text (or viewed a blank screen), we asked them a series of questions.

Some of the outcomes we measured were behavior changes that could directly help animals—for example, interest in signing a petition and intention to reduce fish consumption. We also measured attitude changes that we think may correlate with behavior change based on other research.

We told participants that an animal protection group had launched a petition to address the fish welfare issue they had just read about and then asked how interested they would be in signing that petition. We asked the control group, which hadn’t read about a fish welfare issue, about a petition that addressed fish welfare issues in general.

Each of the six treatment texts about welfare issues made people more likely to sign the petition, and the results were statistically significant. The average increase in interest was around a half point on a 1–5 scale. For example, if people in the control group rated their interest a 3, the treatment group rated their interest a 3.5.

There is some evidence that the text about disease and water quality issues may be more effective at making people willing to sign the petition than the other texts, although this was generally not statistically significant. Aside from that, we were unable to determine how effective the other treatments were relative to one another.

We asked people how they expected their fish consumption to change in the 90 days after the study: Would it increase, decrease, or stay about the same?

The disease and water quality text was the only text with a statistically significant effect, with around 23% of respondents saying they intended to reduce or eliminate their fish consumption, compared to just 12% in the control group.

The slaughter text might also be effective, although the evidence was not statistically significant. The text about crowding appeared to be the least effective at affecting people’s intentions.

Other studies have found that attitudes about farmed animal intelligence and sentience correlate with how much meat a person eats and how open someone is to eating vegetarian. In this study, we measured attitudes by asking people how much they agreed or disagreed with the following statements:

The treatment texts appeared to have no effect on changing people’s attitudes toward fish, with one exception: There was weak evidence that the feeding and handling texts might have had an effect on how much people viewed fish as capable of suffering. Their adjusted p-values were 0.14 and 0.11, respectively, meaning the results were approaching statistical significance. Even if these treatments did have an effect, however, the change would be only about a 0.2-point difference on a 5-point scale, so the effect would be small regardless.

The crowding and breeding texts appeared to be least effective at changing people’s attitudes.

One interesting thing to look at is the control group’s attitudes about fish to get a sense of how people in general view fish.

As you can see, on average people are most likely to view fish as capable of suffering, but least likely to view them as having rich emotional lives.

Research on health interventions has found that people who think they have the power to make a diet or lifestyle change are significantly more likely to do so. In this study, we asked people to state how much they agreed or disagreed with the statement “Eating meals without fish is easy.

None of the treatments had any significant effect on the outcome, and in fact all the treatment texts except slaughter and crowding had negative effects (although these negative effects were also not statistically significant).

Research has found that people are more likely to believe or to do something when they think most other people believe it or do it. In this study, we asked people two questions:

None of the treatments had any significant effect on the outcome.

We will send a follow-up survey to participants after three months to gauge long-term effects of the treatments. In addition to measuring the outcomes listed above, we will measure the following:

There will be an additional write-up of these results.

All treatment texts had a positive effect on people’s interest in signing a petition aimed at addressing the farmed fish welfare issue they read about. There is weak evidence that the disease and water quality treatment text had the largest positive impact on people’s interest in signing a petition.

The disease and water quality text also had the effect of encouraging more people to intend to reduce their fish consumption, and it was the only treatment text to have this effect.

None of the treatment texts had any significant effect on attitudes toward fish, perceived behavioral control (“it’s easy to eat meals without fish), or social norms (“do people think it’s OK to eat fish?).

These results suggest that focusing on the disease and water quality aspect of farmed fish welfare is likely to influence more people to sign petitions and reduce their fish consumption than focusing on any of the other farmed fish welfare issues.

Again, the treatment texts were factual statements about welfare issues and did not discuss any specific qualities of fish sentience or highlight individual fish. We presented the texts the way a scientist or journalist might present the facts, as opposed to the way an animal advocacy group would present the facts. To have done otherwise could have had a meaningful impact on the responses generated. In a subsequent study, we will look at similar questions but with text blocks presented in a style more typical of animal advocacy groups.