Caveats for Poll Data

Political polling is like Sex Panther cologne, 60 percent of the time, it works every time.

Opinion poll data is everywhere.  In a Presidential election year, numbers, numbers, and conflicting numbers are unavoidable.  As you wade through the statistical swamp there are five things you need to consider if you want to maintain solid footing.

#1 – The election of the President of the United States is not a national election.  

The latest Rasmussen poll shows Romney leading Obama by 2 percentage points national.  This is meaningless.  If you want an idea of who will win the election you need to look at each state.  This is elementary, but worth a reminder.

More than one candidate has won the nationwide popular vote but lost the election.  This happened three times, in 1876, 1888, and 2000.  Republicans tend to benefit from this because Democrat voters are highly concentrated in the big urban areas.

Interstate demographic shifts make things worse for the Democrat.  Owing to the recent reapportionment of electoral votes, if Obama carries the same states he won in 2008, he will tally 6 fewer votes in the electoral college.

#2 – Only pay attention to polls that use likely voters.

There are plenty of polls that measure all voters, but they are meaningless.  In a big Presidential election year like 2008 fewer than 60 percent of registered voters actually bothered to vote.  The turnout rate is typically closer to 50 percent.

Different polling outfits will use different criteria for determining who is a likely voter, and some methods are better than others.  But even a flawed method of defining likely voters is better than none at all.

#3 – Democrats are usually overrepresented in polls.

Actually, both Republicans and Democrats are typically overrepresented in the polling sample at the expense of non-affiliated voters.  However, the overrepresentation of Democrats is often more pronounced.

A generic Republican is more likely to vote than a generic Democrat.  That’s just the way it is.  Viewing by specific demographic factors helps explain why.  The poor, the young, and minorities are statistically more likely to vote Democrat and each of these groups is traditionally plagued by low voter turnout.  The youth vote in particular is frequently overestimated.  Of course young people did vote above expectations in 2008, but that was the exception.

#4 – Consider the possibility of Obama bias in polls.

There is an under-studied wildcard in political polling.  Pollsters know that people responding to polling questions have a tendency to be, well, nicer, than they would be in the privacy of the voting booth.   Though they try to assure anonymity, some poll respondents just do not feel confident that their answers are confidential.

It is therefore suspected that some people are not comfortable stating their opposition to the nation’s first black president.  An indication of this tendency is the consistent gap in job approval rating and personal favorability ratings.   In a recent Pew poll Obama had a 44 percent job approval and a 52 percent personal approval.  This gap is larger for Obama than it was for his three predecessors.

#5 – Black pride, White guilt.

This is only a guess.  But it seems plausible to imagine that there were some white voters, particularly among independents and women who voted for Obama because he was black.  Call it a white guilt vote.  Having assuaged their conscience in 2008 these folks may not feel the same.

Again, this is only a guess.  But it also seems plausible to imagine that the novelty of the first black major party nominee drove exceptional turnout among minority populations.  Call it black pride.  If that did occur in 2008, and statistics seem to support it did, has the novelty worn off?  If so, will the minority vote in places like Ohio and Michigan drop to more traditional turnout rates?

There is no way to know if either of these scenarios is real, or even likely, but each could have some potential effect.  Where either of these possibilities offset by an anti-black racist vote?  I do not know, but it seems that if someone voted against Obama because he was black they seem sure to do so again, whereas, someone may have voted for him in 2008 as a way to make history, but may not feel compelled to do so again.

Conclusion

Without even touching on the issues of polling in the age of mobile phone supremacy, the statistics thrown around on the news should be viewed with some suspicion.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s