February 04, 2008

Study: Polls may lack religious conservatives' input

by Andrea Hahn


CARBONDALE, Ill. — When an opinion poll and the actual results of a political primary vote don't coincide, Southern Illinois University Carbondale sociology professor Darren E. Sherkat is not surprised.

Sherkat's recent scholarship combines his interest in quantitative methods and statistics with his interest in religion. His article, "Religion and Survey Non-Response Bias: Toward Explaining the Moral Voter Gap between Surveys and Voting," published recently in the journal "Sociology of Religion," offers an explanation of why opinion polls do not accurately predict actual voting results.

Basically, Sherkat says, many popular opinion polls and surveys do not use scientific methods to ensure their polls are representative of the given population. In addition, the constant demand for immediate, up-to-the-second information encourages the propagation of wrong information.

"We have lost our attention to high-quality data," Sherkat said. "If you have a legitimate answer from last week, that's better than an inaccurate answer today. We'd be better off making predictions with good data."

But one of the most slippery problems is accounting for biases created when particular groups of people refuse to participate in the study.

Sherkat's research indicates that religious affiliation and fundamentalist beliefs often play a role in political survey response – and non-response. In other words, religious factors affect not only an individual's political opinions, but also may affect that person's willingness to respond to a given political opinion survey. In particular, Sherkat said, his study and others recently conducted by researchers at the Pew Research Center have found that political conservatives with strong conservative religious orientation are more likely to refuse participation in political opinion surveys.

"A political opinion survey presented by a university, for example, can be seen by some religious conservatives as a liberal-biased survey because they perceive the university system as liberal-biased," he said. "As a result, they may not respond to the survey."

In turn, Sherkat noted, conservative boycotts of perceived liberal bias can cause those surveys actually to reflect a liberal bias, regardless of the survey-takers' intent. The result, as his research indicates, is that political conservatives in particular are often under-represented in political opinion surveys.

"They feel embattled," he said. "Their response rates are far too low for accurate data in political opinion surveys… Political opinion surveys aren't a random sample of the population – they are at best a random sample of the population who cooperates with the survey."

Sherkat recently studied another group that may feel embattled during political season – atheists and other non-religious people. His views about how non-religious members of American society regard the religious statements often made by political candidates during their campaigns were widely sought both here and abroad earlier in the campaign season.

Members of the media wishing to contact Darren E. Sherkat for his sociological analysis of political campaigning, opinion surveys and actual poll results may do by calling his office at 618/453-7614 or by emailing him at sherkat@siu.edu.