Breaking the Spell: Religion as a Natural Phenomenon

Before we jump in to explain the data, we should ask how sure we are of the assumptions used in gathering them. Just how reliable are the data, and how were they gathered? (Telephone inquiry, in the case of ARIS, not written questionnaire.) What checks were used to avoid biasing context? What other questions were people asked? How long did it take to conduct the interview? And then there are offbeat questions that might have answers that mattered: What had happened in the news on the day the poll was conducted? Did the interviewer have an accent? And so on.⁷ Large-scale surveys are expensive to conduct, and nobody spends thousands of dollars gathering data using a casually designed “instrument” (questionnaire). Much research has been devoted to identifying the sources of bias and artifact in survey research. When should you use a simple yes/no question (and don’t forget to include the important “I don’t know” option), and when should you use a five-point Likert scale (such as the familiar strongly agree, tend to agree, uncertain, tend to disagree, strongly disagree)? When ARIS did its survey in 1990, the first question was: “What is your religion?” In 2001, the question was amended: “What is your religion, if any?” How much of the increase in Non-denominational and No religion was due to the change in wording? Why was the “if any” phrase added?

In the course of writing How We Believe: Science, Skepticism and the Search for God (2nd ed., 2003), Michael Shermer, the director of the Skeptic Society, conducted an ambitious survey of religious convictions. The results are fascinating, in part because they differ so strikingly from the results found in other, similar surveys. Most recent surveys find approximately 90 percent of Americans believe in God—and not just an “essence” God, but a God who answers prayers. In Shermer’s survey, only 64 percent said they believed in God—and 25 percent said they disbelieved in God (p. 79). That’s a huge discrepancy, and it is not due to any simple sampling error (such as sending the questionnaires to known skeptics!).⁸ Shermer speculates that education is the key. His survey asked people to respond in their own words to “an open-ended essay question” explaining why they believed in God:

As it turns out, the people who completed our survey were significantly more educated than the average American, and higher education is associated with lower religiosity. According to the U.S. Census Bureau for 1998, one-quarter of Americans over twenty-five years old have completed their bachelor’s degree, whereas in our sample the corresponding rate was almost two-thirds. (It is hard to say why this was the case, but one possibility is that educated people are more likely to complete a moderately complicated survey.) [p. 79]

But (as my student David Polk pointed out) once self-selection is acknowledged as a serious factor, we should ask the further question: who would take time to fill out such a questionnaire? Probably only those with the strongest beliefs. People who just don’t think religion is important are unlikely to fill out a questionnaire that involves composing answers to questions. Only one out of ten of the people who received the mailed-out survey returned it, a relatively low rate of return, so we can’t draw any interesting conclusions from his 64 percent figure, as he acknowledges (Shermer and Sulloway, in press).⁹

3 What shall we tell the children?

It was the schoolboy who said, “Faith is believing what you know ain’t so.”

—Mark Twain

Â