Friday, April 9, 2010

Subliminal Christianity and Racism

According to the latest post on Hemant's blog,

a study published by Baylor University researchers finds that “Priming Christian Religious Concepts Increases Racial Prejudice” (PDF).

Basically, the researchers presented subjects with subliminal flashes of words, purportedly to test their ability to detect and differentiate between words versus non-word letter groups seen for only a brief period of time (less than 100ms). Some subjects saw neutral words, like butter or house, while others saw words associated with Christianity, such as Christ, faith, Bible and gospel. They then ran the subjects through a battery of situational questions designed to determine their degree of hostility towards the African Americans in the situations.

Kate Shellnutt summarized the conclusions quite nicely on HoustonBelief:

Researchers offer some possible explanations for why these Christian terms have such negative effects. They can cue fundamentalism or political conservatism, which can isolate “out-groups,” or echo the notion of the Protestant work ethic, which has been connected with anti-Black attitudes, the study said.

The researchers didn't draw any larger conclusions from this (i.e., that most Christians are racist or anything quite so extreme), though they did insist it was causal rather than correlational.

The demographics were a little weird:

A total of 73 college students (57 women and 16 men; M = 19.6 years) were recruited from introductory psychology classes to participate in a personality and language usage study. Participants were somewhat ethnically diverse (37 Whites, 13 Asians or Pacific Islanders, 13 Hispanics, and 10 African Americans) but predominantly Protestant or Catholic (n = 43, n = 17, respectively). A few participants were of other religions (Muslim n = 1, Buddhist n = 1, “other” religion n = 8 ) or had no religious group at all (n = 3).

78.1% female? That’s an awfully small male sample size. That's an awfully small overall sample size, for that matter.

However, I think the study is fundamentally flawed. To quote the abstract:

Participants subliminally primed with Christian words displayed more covert racial prejudice against African-Americans (Study 1) and more general negative affect toward African-Americans (Study 2) than did persons primed with neutral words.

The problem is pretty obvious: What’s the overlap? Which of the participants who responded with prejudice and/or negative affect after hearing the Christian words would have responded the same way to the neutral words?

Prior to arrival at the lab, participants were randomly assigned to either the Christian or the neutral prime condition.

This means there was zero overlap. How seriously should we take the claim that, within a specific individual, Christian conceptual language is tied to racial prejudice? There’s no control on an individual basis. They say they pre-screened people on their religiosity and spirituality; what about their possible existing racist tendencies?

Not to mention: They specifically picked African Americans because of the historical racism toward them in America. Couldn’t this create an exaggerated effect?

(On a side note, this concept seems to be a theme for Wade C. Rowatt (one of the researchers). He has been involved in two studies with similar conclusions in the past, according to one of his webpages which mentions that he focuses on “the psychology of religion (e.g., religion and prejudice).”)

Their conclusion:

Consistent with the Christian-racial-prejudice hypothesis, people who were subliminally primed with Christian words reported significantly more covert racial prejudice than did people primed with neutral words. Participants subliminally primed with Christian words did not self-report more cold feelings toward African Americans on the thermometer item than did people primed with neutral words. This experiment reveals an influence of Christian religion on subtle racial prejudice. Priming Christian concepts in American college students caused a slight (but significant) negative shift in attitudes toward African Americans on a covert measure. This effect remained when controlling for preexisting levels of religiosity and spirituality.

My objections:

  1. There was no control for existing racial prejudice, only for religiosity and spirituality.

  2. There was no method used to determine whether priming with Christian words would increase a particular individual’s racial prejudices.

  3. The second experiment was done to “replicate and expand the effects of priming Christian concepts on racial prejudice found in Experiment 1.” However, if you’re going to replicate the effects of an experiment, you don’t modify the experiment – you just repeat it. Otherwise you’re doing a different experiment, and its results have to be taken on their own. For example:

    Participants were asked to complete measures of general negative affect and specific negative emotions (i.e., fear and disgust …) toward African Americans. Including these measures allowed us to determine whether the slight but significant increase in covert racial prejudice observed in Experiment 1 was because of a change in a specific affective or emotional response.

    This is not the same as the original hypothesis: “activation of Christian concepts in Americans increases racial prejudice.” The new experiment tests a hypothesis based on the assumption that the first hypothesis is true, rather than attempting to confirm the original results.
Do these mean that the hypothesis is false? No; I think it's a worthwhile hypothesis to explore, but I think that a much more carefully crafted study would have to be performed. This one just doesn't cut it, in my opinion.

Others on Hemant's blog have pointed out some weaknesses in my objections. For #1:
The very fact that the subjects were randomly assigned to either the neutral-word group or the Christian-word group controls for these two factors. That is, if it were merely a matter of existing racial prejudice, then that prejudice would be just as likely to occur regardless of which words were shown.
I'll concede this. Considering that the study took place in the south, which has historically had a higher baseline of racism, I don't think the degree of pre-existing racism could be the sole decider. The population randomization would also lead to less 'clustering' of people with particular views.

For #2:
By splitting the people randomly into the two groups, and still seeing a measurable difference, doesn’t that clearly indicate the causal effect of the Christian words in individuals? Even without testing anybody more than once, the study shows that, on average, flashing the Christian words will tend to make a typical individual act more racist.
I still disagree with this part. If you only test somebody once, you don't have a baseline for that individual to determine whether or not the priming with Christian words actually did increase their level of racism. You can draw a correlation between higher levels of racism and Christian word priming over the entire sample population, but you can't say with reasonable certainty that it was a causal connection for each member.

I would be interested to see the study repeated with different ideologically tied language and with other ethnic/minority groups as the 'target' of the discrimination. I can understand the idea of ideological language being linked to in-group vs. out-group thinking, but whether subliminal cues can cause an increase in that sort of thinking isn't clear to me. It's worth a further look.