What does “bias” mean?

The Washington Post publicizes a lot of good political science these days. (Thank you, Mr. Bezos!) They recently ran a column entitled, “Americans aren’t biased against Latino immigration. Here’s what they actually fear.” 

Short version: Morris Levy (USC) and Matthew Wright (American) surveyed a bunch of white Californians about their attitude towards immigration. They humanized their questions by asking the respondents about a particular individual, either a Mexican immigrant named “Juan,”  a Chinese immigrant named “Yuan,” or a German immigrant named “Johan.” In addition, half of all respondents were told that the immigrant they read about had lived in the United States for two years. The other half were also told that Juan/Yuan/Johan speaks English and has worked steadily as a waiter.

Here is what they found:

In other words, when there was no information about the immigrant, more white Californians supported legalization when thinking about somebody from Europe or China than from Mexico. But if they were told that the immigrant worked and spoke English, the gap disappeared.

The authors then conclude, “Discrimination against Latinos may grow not from hostility against an ethnic ‘outgroup,’ but rather stereotypes about whether they will contribute to the United States or become a burden.”

It’s a neat result, but I don’t understand the conclusion. They find that when told a male immigrant has assimilated, white Californians don’t care about his ethnic background. But they also find that without any information, Californians assume that Mexicans are much less likely to assimilate than other immigrants. They conclude that people stereotype Mexican immigrants as less educated and more likely to become a public burden. 

That is possible. But there is another explanation. They found that white people in California don’t consider Mexicans impossible to assimilate. But their results are consistent with the hypothesis that white people consider Mexicans to be intrinsically harder for the American melting pot to melt. 

Wouldn’t that make them an ethnic outgroup?

Comments

[Matt McIrvin]

It's a distinction without a difference, it seems to me: this is just how bias works. Racists are often perfectly capable of believing that people of the disfavored group who they personally know are exceptional "good ones". In fact, they brandish the belief as a get-out-of-racism card.

[Dave K.]

Bias implies a prior assumption or an assumption invulnerable to evidence. From a Bayesian point of view, racism is a bias that is insensitive to updating.

The composition of likely Chinese or European immigration is different than for Latin American immigration in terms of education and English language skills. That could inform many people's priors about other undesireable characteristics.

I've noticed the same bias, when more insular Northeasters/Midwesterners know I'm from Alabama prior to meeting me. They rather quickly update their priors when they meet me.

[shah8]

I have to roll my eyes at this.

"Learning the language, willing to assimilate", these are dog whistles for more noxious mentalities, which use "model minorities" to unfavorably compare their targets to. I've certainly heard more than I need to hear of that sort of reasoning. Much as it was with the birth certificate thing, it is usually an act of willful belief without any regard for truth.

Also, I will argue against Dave K. about exactly what people "update their priors" about. Some people don't update their priors about Dave K., but what they can *do* to Dave K.

[Dave K]

shah8, that seems overly cynical. A few years ago I had the duty of arranging ecclesiastical interactions with Ethiopians and their coreligionists in America. I saw quite a bit of bias but I also saw dramatic updating of priors. Beyond my personal anecdotes, the research on personal interactions and persuasion implies many people are adaptable.

[Gareth WIlson]

Figures on real-life two-year immigrants from the three regions would be useful here, right?

[Dave K]

Gareth, I went here for an approximation:

http://www.census.gov/content/dam/Census/library/publications/2014/acs/acs-26.pdf

~1% of German immigrants are English non-proficient, along with 33% of Chinese immigrants and 47% of Mexican immigrants.

Figure 8 (on page 11) shows interesting trends in English speaking proficiency by continent/region of immigration.

[Gareth WIlson]

Thanks.

[Peter]

Interesting, but I'm puzzled by the methodology ... why use German immigrants as one of the examples, when German immigration to the US surely must be almost zero?

[Noel Maurer]

A white nationality not subject to prejudice.

[Dave K]

Peter, depends what you mean by almost zero for German immigration (10,000-15,000 per year since 2000).

Noel, besides Anglo-Canadians is there really a foreign white nationality completely immune to prejudice from certain subsets of America?