Pollster.com

Articles and Analysis

 

Omero: The Palin Effect--Its Rise & Fall


One of the big topics from September was the Palin Effect, and how it improved McCain's standing with white voters, particularly white women.  While most commentators agreed the Palin Effect didn't move Hillary Clinton's primary base, there was some unique movement among white women overall.  And while it's tough to isolate the effect of campaign events after Labor Day (especially given the economic crisis), the post-Palin bounce has been declared over.  We can indeed track its rise and fall.

 

This has been a full team effort.  Mark collected data from 56 different national surveys.  (A spreadsheet with all the datapoints can be found here: White Women.xls.)  And as part of pollster.com's continued upgrade, and Charles Franklin's mighty-fine chart-making, we can look at the presidential vote among white voters, across gender.  There are two charts below; the second is more sensitive to outliers.  But both tell the same story. 

  

 

whitewomen.png

 

moresensitivewhitewomen.pngFirst, it's worth pointing out that much more attention has focused on the Palin Effect on white women than on white men, or really any other group.  Naturally, that's largely due to Palin's gender.  But it's also because white women are a swing group.  Some polls show Obama leading with this group, but our chart shows, on average, McCain leads.  White men have given McCain an advantage in every poll we were able to obtain that broke out results by race and gender.  Gary Langer at ABC discussed the greater variability in white women's vote here, and our spreadsheet shows a larger standard deviation for white women, across all the surveys examined.

 

But despite lopsided attention to and increased variability among white women, there are more similarities than differences in the post-Palin pattern among white men and women.  Among both groups, support for Obama fell.  And with both groups, support for Obama has rebounded.  The difference is in the depth of the fall, and the extent of the rebound.

 

Our charts show white voters across gender moved from Obama, and substantially, in the early days of the Palin bounce (around September 7).  But white women did have a steeper drop.  Our sensitive chart shows white women giving McCain a double-digit lead for a few days (approximately September 8-14), before rebounding to pre-Palin levels.  By contrast, white men had been gradually moving away from Obama for months, and the post-Palin drop was much less steep.  Now Obama's support among white men has rebounded above pre-Palin levels.

 

The race, particularly among white women, will likely continue to be volatile.  But the Palin bounce, and bounce-back, seems to have been replaced by other campaign events.

 

 

Comments
szabe:

It seems worth a mention that Palin was picked a few days before the Republican Convention; did you make any attempt to isolate her impact from that of the convention?

In any event, thanks for the analysis.

____________________

Eternal:

This analysis supports what's been seen in the individual polls. Palin may make the red states deeper red but does not make the toss ups pink.

To be fair, I don't know that it's the VP's job to win states or groups but since some suggested she had/was it's nice to see date that clarifies her impact.

____________________

Observer:

Palin seems to have rallied the base of the Republicans to the McCain ticket. What I have not seen is any analysis of whether she has changed the party ID balance.

All the polls seem to assume bigger or smaller party ID leads for Democrats. I have seen it range from 3 to 9 which of course affects the bottom line a lot. I have also read that it was even in 2004.

If Palin (or the course of the campaign generally) means more people now identify or re-identify as Republicans then maintaining the same party ID percentages will completely mask any sort of Palin effect of bringing back Republicans to the fold.

At its worst polling with a fixed party ID introduces a bias which makes a poll at least party immune to changes in voter preference. Surely some people simply identify with the way they will vote? If you now decide to vote Republican you might well now identify as a Republican. It must be as simple as that for a lot of people, especially that multitude who only vote once every four years.

Yes, I am a McCain supporter and I realise that I might be grasping at straws.


____________________

mmanphd:

Do the sample sizes allow you to determine if there was a Palin bounce among black women?

____________________

Pachete:

Doubtful mmanphd, blacks typically vote 90% for democrats. And the invoking of small town values that is so appealing to the religious social conservatives of the republican party wouldnt have an effect on AA's.

____________________

Margie Omero:

Some responses...

Someone may be able to separate the effect of Palin vs. the GOP convention. But it's my sense that Palin was a bigger campaign event than the convention itself. Especially given the extent and duration of the bounce.

National survey outlets handle party ID differently. I think aggregating the data, and using different levels of sensitivity, accounts for this.

Obama has largely consolidated black vote since the primaries. An unfortunate consequence of this is little public subgroup data on black voters.

____________________

PrinceCyclone:

In fact, there is no evidence whatever suggesting a Palin bounce among white women in these plots. The slide among white women had already begun (in fact, was apparently a week or more old) when she was chosen. There is weak evidence for a small bounce among white *men*, at least in the more sensitive plot, but among white women the most striking thing about where the Palin line falls is that it changes the slope of the trend not at all.

____________________

statsguy:

I think the evidence for the effect on women is there. There's nothing among the pre-pick data points that would suggest that support for McCain is growing among women. The "downward trend" seen in the red curve prior to Palin's pick is only present because of the polling points after her selection, especially the large cluster that appear between -10 and -20 during the first half of September.

____________________

RWCOLE:

Isn't the data equally consistent with the hypothesis that McCain got a normal convention bounce following Obama's campaign bounce- which faded in the normal time- and that Palin had nothing to do with it?

____________________



Post a comment




Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR