An article of faith in a poll
A CBS/NYT poll shows a big rebound for Senator Obama following his recent denunciation of his former pastor, the Reverend Jeremiah Wright. The poll says Obama now leads John McCain 51/41. That would appear to be a big boost to Senator Obama as Indiana and North Carolina go to vote today.
Curiously, however, in a poll in which religious issues are important, the poll appears to undersample religious people. People who attend church at least once a week were 41% of the electorate in both 2000 and 2004. Such people formed only 26% of the sample in the CBS poll. Churchgoers are seemingly underrepresented in a poll about politics and church by over a third. How reliable can the results of such a poll be?
Moreover, people who attend church at least once a week broke 65/35 for George Bush in 2000 and 2004. Undersampling this large and highly relevant demographic group would appear on the surface to call into question the poll’s validity, and make it look like an ad for Senator Obama. Is anyone surprised?
