Dive Deeper Into The Polls

The poll numbers are coming fast and furious now and this will not let up until November 7th so I caution everyone to look a little deeper than the poll headlines before they come to any conclusions.

I use Real Clear Politics to serve as a clearing house of the major polls and you can click on the individual poll to gain further details such as the questions asked, the sample size, margin of error and (this is important!!!) party affiliation of the people polled.

I will use a recent poll from Democracy Corps as an example how you must dive into the details before you glean too much from the overall polling numbers.

From this poll, we see that Obama has a 5 percentage point advantage over Romney with 1,000 likely voters (50% to 45%).  The main question posed was:

“Thinking about the presidential election in November, if the election for president were held today and the candidates were Democrat Barack Obama and Republican Mitt Romney, for whom would you vote?”

Wow, this looks pretty bad for Romney!  But when we look at the party affiliation of the 1,000 people sampled we see a large bias to Democratic voters.  The results below are to the following question:

“Generally speaking, do you think of yourself as a Democrat, a Republican or what?”

Strong Democrats outnumbered strong Republicans (30% to 21%), weak Democrats outnumbered weak Republicans (11% to 9%) and Independents leaning Republican outnumbered Independents leaning Democratic (13% to 9%).  Overall, those affiliated with the Democratic Party outnumbered those affiliated with the Republican Party 50% to 43%.  That is a 7 percentage point delta in a poll that suggests Obama has a 5 percentage point lead over Romney and this poll has a 3.1% margin of error.

What if the poll were able to sample equal number of people who were affiliated/leaning Democrat versus Republican?  For fun, let’s assume that those who categorized themselves as Strong, Weak or Leaning to a party would vote for the Presidential candidate of that party.  If we were to theorize that the Democracy Corps polling sample had equal percentages of Strong/Weak/Leaning party affiliation then we would see that Romney would lead Obama by 1 percentage point.

The calculations can be seen in the table below where I redistributed the 9% delta in Strong Dem, 2% delta in Weak Dem and 4% delta in Independent Leaning Republican equally amongst the parties.

But even leveling the party affiliations is still a poor representation of how people will actually vote.  The following data from Rasmussen show that the advantage of those affiliated with the Republican Party surpasses those affiliated with the Democratic Party anywhere from 0.9 to 4.3 percentage points in 2012.

August 31, 2012
Republican Democrat Other R – D Quarterly
2012
Aug 37.6% 33.3% 29.2% 4.3%
Jul 34.9% 34.0% 31.1% 0.9%
Jun 35.4% 34.0% 30.5% 1.4% 1.8%
May 35.7% 33.8% 30.5% 1.9%
Apr 35.1% 33.1% 31.8% 2.0%
Mar 36.4% 33.4% 30.2% 3.0% 3.3%
Feb 36.0% 32.4% 31.6% 3.6%
Jan 35.9% 32.5% 31.6% 3.4%

So keep in mind that any polling data that is either skewed to the Democratic side or has equal amounts of Democrats and Republicans is biased and doesn’t reflect voting reality.  As the polling data is released and the Leftists start thumping their chests, bring them back to Earth by diving deeper into the data.

This entry was posted in politics. Bookmark the permalink.

3 Responses to Dive Deeper Into The Polls

  1. tomwys says:

    Worth digging up are the poll results two weeks before the Carter/Reagan (or Reagan/Carter; no bias here!) election.

  2. Pingback: Post Debate Bounce | cosmoscon

  3. Pingback: Romney Took A Knee | cosmoscon

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 )

Google photo

You are commenting using your Google 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 )

Connecting to %s