By allowing ads to appear on this site, you support the local businesses who, in turn, support great journalism.
Divine knowledge: The unique challenge of religion in the polling industry
Photo illustration by Aaron Thorup - photo by Chandra Johnson
When Amanda McAvoy recalls her childhood, one of her favorite memories is praying as a family before dinner.

Its so ingrained I still do it, the Nebraska mom said. My husband and my daughters dont know I do it, but I just take a quick second over my plate and say that prayer in my head.

Growing up in the Midwest, it was second nature for McAvoy's family to pray before meals and attend their Lutheran church regularly. But during her college years, she fell out of the habit of going to church. Since having two children in the past couple of years with her husband, McAvoy, 33, says shes trying to figure out how faith would best fit into her new family.

I dont doubt God, I dont doubt Christ, McAvoy said. But I dont know that what I believe fits into this box marked Lutheran and thats that.

McAvoy is what statisticians might call a None that is, the increasing number of Americans who say they are have little or no religious affiliation and the number of people like her is a subject of debate among religion researchers.

According to Pew Research Center, whose landmark research on the None trend coined the term, the number of religiously unaffiliated Americans is significant: In 2012, Pew found that they made up one-fifth of the U.S. public and one-third of adults under age 30.

But Robert Wuthnow, Princeton University Center for the Study of Religion director, says what we know or think we know about "Nones" and other reported religious trends could be incorrect.

That's the controversial claim of a new book by Wuthnow that's making waves among those who study religious trends.

In his book Inventing American Religion: Polls, Survey and the Tenuous Quest for a Nations Faith, Wuthnow argues that the number of the unaffiliated may seem more dramatic if the baseline data that modern polling firms rely on isnt accurate in the first place. Wuthnow argues that many statistics about religion in America are so inaccurate that they would be disregarded in other subject areas such as political rankings or vote tallies. Church attendance rates are a simple example, Wuthnow said.

The best estimates are that when pollsters tell us that 90-100 million people attend church on Sunday, theyre off by 30 million. Thats scandalous, Wuthnow said. If pollsters can figure out how many people are voting, they should be able to figure out how many people go to church.

Yet to large religion polling firms like Pew and Barna Group, the questions Wuthnow raises are not new. Its part of an ongoing struggle to get accurate results amid significant technological and cultural changes, which most polling firms argue they successfully mitigate.

Theres a circular argument when you say what we know today isnt true because nothing is accurate, because youre calling into question the very evidence you have that the numbers are inaccurate, Barna president David Kinnaman said. Polling isnt going anywhere. Its a major part of our world because it works.

A superficial portrait?

In his book, Wuthnow argues that polling's dependence on the news media led to a rough sketch of religion that lacked depth.

From the beginning, polling was in the business to make headlines, and that is pretty much what it continues to do today, Wuthnow recently wrote for First Things. Covering whatever topics happened to be in the news, pollsters rarely brought in-depth knowledge of religion to the questions they asked about religion.

Wuthnow attests that because polling was often conducted for and relied upon by newspaper and broadcast media coverage, questions concerning religion were often ignored by news outlets as not newsworthy or oversimplified to avoid confusing readers. This oversimplification has persisted for decades, Wuthnow argues, and has painted a superficial portrait of religion in America that people have accepted as truth elevating pollsters as very powerful interpreters of American religion.

Many researchers agree that questions are written to apply and make sense to as many people as possible and poll results are often oversimplified when reported in the media. But they also contend that polls were only meant to serve as one piece of a larger understanding of religion through research.

Its indisputable that polling can answer some kinds of questions, but not every kind of question about religion. Weve never asserted otherwise, Pew associate director of research Gregory Smith said. But for anyone to be left with the impression that we should assume that we cant learn anything from religion polling would be a shame.

Even with its limitations, religion researchers say that polling provides a unique service to a curious public.

Is it possible to completely capture the experience of subgroups like African-Americans or Catholics by asking them a few questions on a survey? Of course not, said Dan Cox, research director for the Public Religion Research Institute (PRRI). But polling helps people understand other perspectives, and theres value in that.

Without religion polling, Kinnaman says faith leaders, policy makers and even business people hoping to create goods and services for religious groups would be flying in the dark.

Polling is just part of the pantheon of what we know its not the best way and its not the worst way to learn about religion, Kinnaman said. Its easy to take the facts and knowledge we have now for granted, but if we didnt have polling, we wouldnt have a map for religion and wed have no descriptions to go on.

While Wuthnow doesnt call for an end to polling, he does wonder how useful the current map of the religious landscape is.

If the devil is in the details, Wuthnow wrote, the details about religion polls are devilishly difficult to trust.

Quantifying faith

Some of the problems Wuthnow attributes to inaccuracies in religious polling data are issues the polling industry faces more generally, such as the fact that fewer people are taking polls than ever.

With answering machines, caller ID and the decline of landline phone use, response rates have fallen from about 36 percent in 1997 to below 10 percent in some cases today.

As Rutgers University political scientist Cliff Zukin wrote in the New York Times this month, only 6 percent of Americans were estimated to use cell phones instead of landlines in 2005. By the first half of 2014, that figure jumped to 43 percent.

Wuthnow argues that declining response rates make the pool of people who respond so small that it can bias poll data.

But Smith of Pew and Kinnaman of Barna Group say pollsters have worked hard to mollify these problems by ensuring their samples are representative of the groups they're studying or the population as a whole. Pew's own study into the impact of plummeting response rates found that although sample sizes were smaller, the firm took measures to make sure the samples were still representative.

Its an important issue and its an issue we (at Pew) have spent a great deal of time monitoring, Smith said. But its not clear to me that theres inaccuracy of such magnitude that we should be calling the trends into question.

Quantifying a person's faith poses unique challenges to researchers beyond getting enough people to take a phone survey. For one thing, researchers have no historic context for religious belief or affiliation because the U.S. Census does not ask questions about religion or spiritual belief, said University of Akron political scientist and Pew senior religion poling fellow John C. Green.

If we were measuring gender, for example, we have a good fix on how many men and women there are in the U.S. because of the census. With religion, we have no such measure, Green said. It makes these surveys particularly important, but on the other hand, it also makes them uniquely controversial because we dont know what the real numbers are.

New York University political scientist Jeanne Zaino says trying to measure how religious a person is wholly different from asking them about their feelings on political issues.

Religion is intensely personal and because its so sensitive and controversial to talk about in some cases, theres a huge potential for people to either not want to respond or to respond in an idealized way, Zaino said. Are they telling you what they think you want to hear, or the truth? With religion, thats hard to get to.

Difficult or not, Wuthnow argues that its precisely because religion is a sensitive and complex subject that pollsters should work harder to improve their methods. Simply asking additional questions, he said, could add depth and information to results something Wuthnow says academic polling does.

It gives an aura of precision to say that Pew and Gallup show 90 percent of Americans believe in God, but the numbers say nothing about what people believe about God or how they experience God in their lives, Wuthnow wrote. The question is whether the challenges are so great that polls will or should become a relic of the past.

To that, Kinnaman unequivocally says the answer is no. Despite its restrictions, religion polling holds an important place in understanding American religion.

The polling industry has worked hard to create a more accurate picture of what we can understand through research, Kinnaman said. A world without religion polling would be a world without a spiritual map for what people believe. It helps people name and describe and give context to the human experience.
Sign up for our E-Newsletters