Home 2013 races Polling Puzzle: The Missing Evangelical/Born-Again Voters

Polling Puzzle: The Missing Evangelical/Born-Again Voters

187
0
SHARE

by Paul Goldman

In the 2009 Virginia gubernatorial election exit poll, “White Evangelical/Born-Again” voters made up 34% of the electorate, going 6:1 for winner Bob McDonnell (Creigh Deeds, who lost by a landslide, actually carried the remaining voters by a bare majority). However, in the latest 2013 Virginia governor’s race, the respected Marist/NBC poll says “Evangelic Christian[s]” will only constitute 19% of the electorate (I used Marist, since conservatives think the Post Poll is biased against them, and the Q-poll doesn’t seem to have asked about religion).

A few months ago, here on Blue Virginia, I caught “hell,” as John Wayne might have said, when writing that white Evangelical/Born-Again voters were likely to be the key voting group this year if Cuccinelli ran the kind of campaign myself and others expected. But facts are facts: Cuccinelli, in my view, needed a McDonnell vs. Deeds type of electorate to have a chance of winning, and this didn’t seem remotely possible to me even though the GOP guy led by 11 percentage points in the Washington Post poll back then.

In 2012, the exit pols say that “White Born-Again Christian” voters made up 23% of the electorate. They voted 6:1 for Romney. Thus, my polling puzzlement: How is it logically derived that “Evangelical” voters can only make up 19% of the Marist projected electorate?  Notice I used the word “logically.” Anything is possible in politics. But the Marist 5 percentage-point Terry margin and its 3-point Dem over GOP projected electorate are both consistent with other polls, thus raising a fair question: how is the 19% evangelical share logically derived, given the hard data in recent years?

The Marist poll has white voters at 72%, and the 2009 exit polls had it at 78%. African-American turnout is pegged at 19% this year, compared to 19% in 2012. Accordingly, you see my polling puzzlement:  Where did the “White Evangelical/Born-Again Christian” vote go?

 

Logically, the one thing E.W. Jackson should in theory be able to do in a positive way for Cuccinelli is to rally this voter group. What more does the GOP LG have to do? Unless, of course, as I believe, Mr. Jackson is such an embarrassment to himself and others that folks naturally don’t even want to be associated with his stuff. Perhaps people figure Jackson has to be auditioning for a role as Ann Coulter’s co-host on the new Fox show “Crazy for You,” with the hit record of the same title as the theme song.

Still, dropping from 34% all the way to 19% when it was 23% in a big Democratic year in 2012 here in Virginia? It raises the question: since these voters broke 6:1 for McDonnell and Romney (McCain too if I remember), what is the net/net effect if the polls are significantly understating this group? Also, why is this happening?

As a statistical matter, it might not be a big deal at all, given the partisan nature of the 2013 electorate, or it might be a big deal depending on weighting and other techniques used by pollsters. Indeed, the Marist poll underweights African-American voters, a key Democratic group. This could mean a skew against McAuliffe, depending on other assumptions, or effectively a wash depending on your other data markers.

But the BOTTOM LINE: Clearly the Marist Poll, in surveying voters, is not picking up the kind of White Evangelical Born-Again voting propensity as would have been logically assumed based on recent statewide elections.  Otherwise, they would not be using the 19% figure for a group that made up 34% of the electorate in 2009 and 23% in 2012 when the Obama constituency came out in strong force.

My take: The Cuccinelli camp is strongly in the Karl Rove/Dick Morris 2012/Romney camp mindset, believing the published polls are dead wrong on turnout and that their polls, weighted far more “correctly,” show a Cuccinelli/Jackson/Obenshain sweep. Indeed, they probably share a private laugh every morning chuckling at how the VA Dems are going to be shocked on election day when a huge GOP voter posse, hiding in the “polling weeds,” rises up and swamps the polls.

My take? The Republicans are engaging in gallows humor for sure, but it’s their funeral, so I will not interfere. Yet at the same time, a serious observer trying to be fair has to ask one last time: Why are polls like Marist assuming such a low turnout from the strongest part of the GOP voter base? Logic would suggest these people might be itching to vote to send a message to LG Bolling and the others who have deserted the party for personal gain. Or, is there a different message if this strong GOP group intends to sit out the GUV race in what would be record numbers?

Like I say, it is a polling puzzlement for sure: White Evangelical Born-Again Christian voters missing in action, because it would ALSO SUGGEST that the Tea Party folks are likewise going to stay home, despite Cuccinelli and Jackson being “tea party before it was cool” (in GOP circles, at least, although losing coolness at a record rate if you look at the polls).

BOTTOM LINE: As I wrote months ago, “34%” is a bridge too far. But dropping all the way down to “19%” is a meltdown that is sending quite a message if true. That being said: The Marist Poll only had a 3 percentage point Democratic margin over Republicans in the electorate, and it was just 4 percentage points GOP over Dem in 2009. Thus, logic suggests the poll is picking up strong Republicans who may be Evangelical voters but who didn’t respond to certain survey questions. And again: The statistics of polling assumes various counter-balancing variances on each side, the cumulative pluses and minuses roughly offsetting 19 out of 20 times. Sub-group data therefore is important to get right, but as long as mistakes are random, you get a good poll.  

Whatever the explanation, it seems to me that the pollsters should clarify the puzzle in their next round of polling before the conspiracy theorists take another piece out of the diminishing credibility of the political process. There is no useful good to such a situation.