Sunday, November 11, 2018

Contact hypothesis bites the dust. Or at least it takes a big stumble.

From The contact hypothesis re-evaluated by Elizabeth Levy Paluck, Seth A. Green and Donald P. Green. From the Abstract:
This paper evaluates the state of contact hypothesis research from a policy perspective. Building on Pettigrew and Tropp’s (2006) influential metaanalysis, we assemble all intergroup contact studies that feature random assignment and delayed outcome measures, of which there are 27 in total, nearly two-thirds of which were published following the original review. We find the evidence from this updated dataset to be consistent with Pettigrew and Tropp’s (2006) conclusion that contact “typically reduces prejudice.” At the same time, our meta-analysis suggests that contact’s effects vary, with interventions directed at ethnic or racial prejudice generating substantially weaker effects. Moreover, our inventory of relevant studies reveals important gaps, most notably the absence of studies addressing adults’ racial or ethnic prejudices, an important limitation for both theory and policy. We also call attention to the lack of research that systematically investigates the scope conditions suggested by Allport (1954) under which contact is most influential. We conclude that these gaps in contact research must be addressed empirically before this hypothesis can reliably guide policy.
Contact Hypothesis has been a naive article of faith in social psychology circles for some decades. A bit of a Kumbaya, can't we all be friends" kind of faith. From Wikipedia
In criminology, psychology, and sociology, the intergroup contact theory has been described as one of the best ways to improve relations among groups that are experiencing conflict. Gordon W. Allport (1954) is often credited with the development of the contact hypothesis. The premise of Allport's theory states that under appropriate conditions interpersonal contact is one of the most effective ways to reduce prejudice between majority and minority group members. If one has the opportunity to communicate with others, they are able to understand and appreciate different points of views involving their way of life. As a result of new appreciation and understanding, prejudice should diminish. Issues of stereotyping, prejudice, and discrimination are commonly occurring issues between rival groups. Allport's proposal was that properly managed contact between the groups should reduce these problems and lead to better interactions.
Pragmatically, this has meant a lot of social policies predicated on the belief that if we just expose rival groups to one another, they will realize we are all just folks and end up being friends.

Except, that's no what the evidence shows. The contact hypothesis only works if 1) people are open and good willed, 2) prejudices are based on blatant false knowledge, and 3) people reason only from direct experience and not the experience of their affiliative groups. These three conditions are rarely satisfied.

As with any controversial and less than obvious naivety out of academia, the idea lives on owing to an excess of poorly designed studies.

Lee Jussim has a pretty good summary of the above research.



Extracting the thread:
The Contact Hypothesis is a Mess: Thread

The Contact Hypothesis (hence, "CH") is an old idea in social psych: That contact between groups reduces prejudice. This was one central theme of Allport's 1954 classic The Nature of Prejudice, built on even earlier ideas.

Beneficial effects of contact have always been difficult to obtain, requiring an ever-growing list of conditions supposedly necessary or at least beneficial to get it to work (equal status, cooperation, common goals, supports from authorities, and more).

This recent award-receiving meta-analysis by Pettigrew & Tropp (2006) gave the answer so many social psychologists had been (I suspect) rooting** for. Slam dunk, contact works!
** nearly 6k citations
** supports left view/values (eg, immigration? let em all in, contact works!).

Except ... there was always reason to doubt this. In the real world, Putnam showed at about the same time, in both work groups and communities, diversity lowered cohesion & social trust, and led to high turnover and lower public investiment.
https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1467-9477.2007.00176.x

A 2014 updating by van der Meer & Tolsma (ht @Chrismartin76 ) found basically the same thing, especially in the U.S.
https://www.annualreviews.org/doi/abs/10.1146/annurev-soc-071913-043309?journalCode=soc

WTF is going on? No one really knows, but enter @betsylevyp and her team, with this amazing 2018 paper:
http://ow.ly/y8qW30myb88

They conducted their own meta-analysis, starting w/ALL the studies in Pettigrew&Tropp's. But to be included, they required studies to meet all of the following criteria:
1. They had to randomly assign people to contact. I.e., they only included true experiments, which is the clearest way to eliminate correlation&causality inference problems.

2. They had to measure intergroup outcomes more than one day after the treatment. That is, there had to be at least some evidence that the effect was not completely fleeting and ephemeral.

3. The studies had to have actual face-to-face contact.

4. There had to be a no contact control group.
P&G had 713 samples from 515 papers.
By the time Paluck et al's (some might argue, "minimalist") standards were met, there were 8 papers reporting 9 studies.

They then scoured the lit for studies meeting their standards post 2006. They found a bunch, bringing the total up to 27 studies (still a far far far far cry from the 713 of P&G).

Here are their main findings, reported in Figure 1 of p. 18 of their paper. Its hard to make out, but you have the link to the actual paper.

Click to enlarge.

Several patterns are notable:
1. The effects hover near 0.

2. The one exception is for contact w/ppl w/disabilities. Remove that, and the results are still above 0 (ie, *some* effect of conctact), BUT:

3. Fig 1 plots the effect size against the std errors (SEs). Smaller Ns produce larger SEs, and Fig 1 shows larger effects w/larger SEs (smaller samples). This is classic evid. of publication bias.

Note also the sloping line. It means the larger the sample, the smaller the effect. In fact, when they used SE to predict effect, the intercept was negative, meaning that the *predicted* effect of large samples (low SEs) is to (slightly) *increase* prejudice, not reduce it.

4. Interestingly, the effects for the groups social psychologists seem to be generally most concerned about -- groups oppressed based on race, religion, sexual orientation -- the effects hover barely above 0, especially for larger studies.
Bottom lines? We know a lot less about contact than Pettigrew & Tropp's meta-analysis has led us to believe. There may be a there there, but if there is, that there is a helluva lot less and more equivocal than the there that is cracked up to be there.

Contact almost surely can be either harmful or beneficial with respect to intergroup hostility. But, just as surely, the benefits of contact have been wildly oversold to an overeager social psychology consuming audience. End.
Sixty-two years on from the original articulation of the hypothesis, we still don't have a clue as to whether the Contact Hypothesis has validity and yet it is accepted as true up and down the Mandarin class, often with negative consequences to the populace-at-large.

The social disaster of desegregation busing, now nudged beyond the Overton window, was in many ways founded on a naive understanding of Allport's contact hypothesis.

Plauck et al, in their conclusion, emphasize that Allcock was much more subtle in his argument than the enthusiastic interpretation by the Mandarin class. Allcock was skeptical that mere intergroup exposure would make any difference and indeed might exacerbate hostilities. Allcock specified that the contact hypothesis might be operative under very particular circumstances, hostility and prejudice
may be reduced by equal status contact between majority and minority groups in the pursuit of common goals. The effect is greatly enhanced if this contact is sanctioned by institutional supports (i.e., by law, custom, or local atmosphere), and provided it is of a sort that leads to the perception of common interests and common humanity between members of the two groups.
So Allport's specifications for the Contact Hypothesis were that it might work when
1) The parties interact on a basis of equal status.

2) They have a common goal.

3) The contact is sanctioned by institutions and authorities respected by the parties.

4) The contact creates a recognition of common interests and common humanity among the parties.
What Plauck et al discovered is that this sophisticated rendition of the Contact Hypothesis has hardly been tested at all.

And for anyone involved in international diplomacy and/or negotiations, these are pretty bog standard preconditions.

Yet the Mandarin class for decades has been assuming that all intergroup prejudices and hostility arises simply from ignorance. If you mix them up, it will all be fine.

But the history of mankind, and diplomacy and decision-making and project management suggest otherwise. Most initiatives fail and they fail for predictable reasons such as misaligned goals, different problem definitions, disparities in power, access to resources, different prioritizations, etc.

As Jussim said, What a mess. It would seem like academia and the Mandarin class would demonstrate some humility and show the self-discipline not to impose their untested theories on others until they actually know what they are talking about. But that is not their way. If it is emotionally appealing, they impose it through public policy until it provess too disastrous and then slink away, the destruction and carnage borne by others.

No comments:

Post a Comment