Change minds, change the world!

I’ve often wondered why people want less immigration. Could it be that they believe that immigration will hurt their job prospects or wages? Put too much of a strain on public services? Commit crimes, or change our culture or politics for the worse? 

If you think the sort of immigration rates we’ve had in Britain since the mid-2000s are good, as I do, you want to change the minds of people who don’t. Hell, you want to know if it’s even possible to change their minds. Maybe arguments and ideas don’t really change anything.

I’ve blogged before about a working paper that may show that you can change people’s minds:

We observe that participants in the treatment group update their beliefs about immigrants, and they donate more money to a pro-immigrant charity. These effects are fairly large, and they correspond to a change of approximately 0.25 of a standard deviation. Moreover, these effects persist even four weeks after the treatment. 

However, participants who receive the information treatment do not become more supportive of immigration reform. Indeed, they do not become more willing to sign a petition in favor of immigration reform, and their self-reported policy preferences remain broadly unchanged. Still, they are less likely to state that there are too many immigrants in the U.S.

Furthermore, we find that Republicans respond more strongly to the information treatment, both in terms of their views on immigrants and in terms of their policy preferences. Indeed, Republicans who receive the treatment become more likely to report having signed the petition in favor of immigration, and they become generally more supportive of immigration. Similarly, we observe that people who are initially more worried about immigration react more strongly to the information we provide them, and they update their views on immigrants and immigration more drastically than people who are less worried about immigration.

From Japan, with quite strict restrictions on immigration, we have a paper that finds that information campaigns about the potential socio-economic benefits that come from immigration do shift people’s views quite a lot:

Depending on the treatment, we find that this exposure led to increased support for allowing more immigrants into the country by 12-21 percentage points, or over 70% above the baseline rate. The treatments also motivated citizens to take political action in support of a more open immigration policy. Notably, while smaller in magnitude, many effects also persisted 10-12 days after the treatment.

From Norway, a Master’s thesis based on a randomized experiment investigating ‘framing effects’ (a concept I don’t think is very useful, but never mind) around whether people are told that ‘just’ 60 percent of immigrants are in employment, or that 7 percent are unemployed. (Others may be students, children or retirees.) Does positive-sounding information move people more or less than negative-sounding information? 

In my first hypothesis, 𝐻𝐻1, I posited that it is more likely to find statistical significance for the negative framing than it is to find statistical significance for the positive framing. I found support for this hypothesis in both my regressions, and phenomena of loss aversion, “losses loom larger than gains,” and a negativity bias may explain the results. Other explanations include the fiscal burden hypothesis, that people fear higher taxes or lower benefits, and social identity theory (Tajfel & Turner, 1986). In the latter case, by accentuating certain features of immigrants (work status/race/origin) in a frame, one reminds respondents of the out-group status of the immigrants. This reminder strengthens the in-group mentality, and a negative frame may thus strengthen the disfavor of the out-group.

In 𝐻𝐻2, I asked if it is more likely to find statistical significance for the negative impact framing than it is to find statistical significance for the negative behavioral framing. I found that for views on the cost/benefit of immigration, both treatments were statistically significant, though negative impact framing (p<0.01) more than negative behavioral framing (p<0.05), supporting the hypothesis. However, this variable pronounced weaknesses of experimenter demand effects. 

For immigration liberals like me, this is a sign of the importance of challenging negative untruths about immigrants.

And most recently, a paper by one of the same authors as the first study. This finds that if you do change people’s beliefs about how immigrants affect the labour market (ie, they typically do not hurt natives’ wages or job prospects) with, in this case, information about the Mariel Boatlift, you can substantially shift their views about immigration policy:

we find that a one standard deviation change in beliefs about the economic impact of immigration changes attitudes towards allowing more immigration by between 0.5 and 0.6 of a standard deviations. Through the use of real online petitions, we also find that changes in attitudes affects real political behavior. Finally, we find that the effects persist in an obfuscated follow-up study where differential demand effects across the treatment and control group are of no concern. Overall, these findings challenge the current consensus that labor market concerns are not a quantitatively important determinant for immigration attitudes.

These studies all seem to back the ‘naive view’ of opposition to immigration, which is that it is motivated by what people tell us it’s motivated by– like jobs, wages and welfare – and not the view that people are actually concerned about racial or cultural change but are afraid to admit that to a pollster. And they’re evidence that honest debate is worth engaging in if you want to change people’s minds and public policy. Nice one!