Lord, C., Ross, L., & Lepper, M., Biased Assimilation and Attitude Polarization: The effects of Prior Theories on Subsequently Considered Evidence. JPSP, 1979, 37, 2098-2109.

"People who hold strong opinions on complex social issues are likely to examine relevant empirical evidence in a biased manner. They are apt to accept "confirming" evidence at face value while sujecting "disconfirming" evidence to critical evaluation, and as a result to draw undue support for their initial position from mixed or random empirical findings".

Thus the result of exposure to a group holding different views can be polarization This result was confirmed using two variations of studies on the death penalty.

Often opinions on important social issues often "survive strenuous attempts at resolution through discussion and persuation". However data relevant to a belief are not processes impartially. People accept the data that suports their initial beliefs and rejects the information that doesn't support it, leaving them even more convinced of their opinions. It's called a "biased assimilation process". People can:
* remember strenghts of confirming data but weaknesses of disconfirming data
* judge confirming data reliable and disconfirmaining data unreliable

Procedure
Based on questionnaire information, they selected 24 proponents and 24 opponents to the death penality (undergraduate psych students). In groups they individually read cards with either confirming or disconfirming results of studies about the death penality. They read the cards and then answered questions about attitude change. Then they read the full procedures and results of each study and then judged how through and accurate each study was. Then they answered attitude change quesitons again. They repeated the procedure with another study with opposite results.

Results
There was strong bias toward the report that matched their initial opinion on the death penalty. There was also more polarization in attitude at the end of the experiment.

Interestingly, after exposure to only one study, both groups move toward the findings from the data, but they more affected those who were already predisposed to the position. When they reviewed the full studies, they tended to ignore the final results and found data within each study (either confirming data or disconfirming flaws in the experiment) to support their initial beliefs. Any initial swaying at first by just reviewing the results was eliminated when they examined the entire procedure.

This shows how people can find pieces of data in any study to bolster and continue their views, even when the bulk of the information is to the contrary.

Other studies confirm that people can persist in their beliefs even when all of the evidence is discredited.

It shows that social scientists should not expect that rationality, elightenment and consensus to emerge when they publish new "conclusive" data about an important social issue.