Faith Matters Survey, 2011

Data Archive > U.S. Surveys > General Population > National > Other > Summary


The 2011 Faith Matters Survey was conducted on behalf of Harvard University and the University of Notre Dame by Social Science Research Solutions/SSRS. The survey was generously funded by the John Templeton Foundation. This collection reinterviewed the respondents from 2006 Faith Matters Survey and also surveyed a new sample of respondents, asking questions about their religion (beliefs, belonging and behavior) and their social and political engagement. The data provide precise measurements of religious belief and behavior to help scholars determine their relative stability among different sub-populations and as compared to nonreligious beliefs and behaviors. Some variable names have been modified by the ARDA. Original variable names are in parentheses.

Data File
Cases: 4,069
Variables: 489
Weight Variable: WEIGHT07
Data Collection
Date Collected: Recontacts: March 3 - June 13, 2011 and August 26 - September 12, 2011 New General Population sample: April 1 - June 12, 2011 New Youth sample: April 14 - June 12, 2011
Original Survey (Instrument)
Original survey website
Funded By
John Templeton Foundation
Collection Procedures
For participants from the 2006 Faith Matters Survey who were recontacted to participate in the 2011 survey, data were collected over two stages. In the first stage, respondents were surveyed by phone. In the second stage, respondents were surveyed both over the phone and online. For newly sampled respondents, data were collected via phone.
Sampling Procedures
Participants from the 2006 Faith Matters Survey were recontacted to participate in this survey, and investigators were able to reach over half of them. Of those who were recontacted, 93% participated in the 2011 survey. The overall reinterview rate is 54%.

In addition to the respondents who were recontacted, two new samples were drawn for the 2011 data. The first new sample is generalizable to the general American adult population, and the second one consists of youth (ages 18-29). The two new samples were selected using a dual-frame landline and cell phone RDD design.
Principal Investigators
Robert D. Putnam, Harvard University
Thomas Sander, Harvard University
David E. Campbell, University of Notre Dame
Related Publications
Putnam, Robert D. and David E. Campbell. 2010. American Grace: How Religion Divides and Unites Us. New York: Simon & Schuster.
Weighting and Estimation
The data are weighted by gender, age, race, region, and education. Qbal, an industry-standard weighting program that employs a raking procedure to simultaneously equate weights by each variable, was used to create the final weights. Counts for the weights were obtained through the 2009 American Community Survey. The weights also account for the probability that respondents from the 2006 Faith Matters Survey could be recontacted.
Exerpt from the epilogue to American Grace
Roughly one year after this book was first published, and six years after our first Faith Matters survey, we returned to as many as we could track down of the people who had spoken with us in 2006 about their faith and civic life. In the intervening years, some of our original respondents had died, and many more had moved beyond reach. But we were able to track down about 1,810 of them, somewhat more than half of the original 3,108 respondents—a quite respectable result for panel surveys of this sort. Of those whom we reached, almost everyone (93 percent) agreed to speak with us again. In addition, we interviewed an entirely new sample of 961 Americans. With appropriate weighting of these new respondents, plus the previously interviewed 1,685, we thus have a total sample of 2,646 in 2011, whose views in the aggregate can be compared with our original 3,108 respondents from 2006. Nestled within these larger samples is a substantial group of 1,685 Americans of all ages and walks of life whose individual views and behavior we can track over these five tumultuous years. Careful analysis confirms that the resulting interviews give us a high-fidelity overview of Americans’ religious and civic attitudes and involvements in 2006 and in 2011 and how those changed, both collectively and individually.
A Note on Response Rates, from American Grace
Our overall reinterview rate of 54 percent compares favorably with the reinterview rate of 62 percent recorded in a similar panel study during these years under the auspices of the General Social Survey, widely acknowledged as the “gold standard” in American academic polling. In any panel study it is harder to track down people who move frequently, and for that reason panel mortality (the “drop-out” rate) is higher for younger people and minorities. That purely demographic bias can be corrected by proper weighting, and with or without weighting we find virtually no evidence that panel bias has influenced any of our findings. For example, highly religious people were no more and no less likely to have dropped out of the panel than highly secular people. We have examined in great detail whether those who dropped out of the panel had differed in any way in 2006 from those who stayed in the panel in 2011, but except for the obvious and correctible under-representation of young people and minorities, we find no significant panel bias. Because the 2006 sample obviously did not include any young people who would come of age between 2006 and 2011, we designed our 2011 sample to over-sample that age cohort, so that the full 2011 sample (properly weighted) accurately reflects the full adult population in 2011, including those too young to have been interviewed in 2006. Finally, whenever possible we have compared our results with the near-simultaneous panel study in 2006-2008-2010 in the General Social Survey, and in every significant case our findings are replicated in those data (though, of course, we have many measures of religious beliefs and behavior that are not assessed in the General Social Survey). As the original edition of American Grace explained, we also reinterviewed many of our 2006 respondents again in 2007, so for many of our respondents we actually have three waves of interviews—2006, 2007, and 2011. In subsequent publications we will exploit all three waves, allowing dynamic analysis of great power, but in this epilogue we focus on the simple comparison of 2006 and 2011.

Bookmark and Share