Friday, October 17

Still Thinking About Polls

Despite the prior two posts I've done in two days on polls, I don't usually pay a ton of attention to them. But my interest was piqued today on Yahoo.com by a headline I've seen all day, "Poll: Voters Souring on McCain, Obama Stays Steady." To be honest, I thought voters had soured on McCain about a week ago so I was expecting to see an even larger dip when I clicked over to the article.

The article is a couple of pages long and goes through all kinds of questions that were posed to voters on whether they find the candidates likable, experienced, honest, compassionate, inspiring, attractive, ethical, etc. On those questions, the poll shows that favorable reaction to McCain has trended downward since September. For Obama, his favorable results have trended upward.

The question I wanted answered was "who would you vote for if the election was today?" Well, if you read the article you won't find that answer. You walk away concluding that probably Obama has widened his lead on McCain based on the results given.

However, I noticed they had a link to a PDF file for the full poll results. If you go to page 6, it says the question was asked, "If the 2008 general election for President were being held today and these were the candidates, who would you vote for?"

According to the full poll results, Obama only leads McCain by 3 percentage points (39-42), with a 3 percent margin of error.

Another interesting fact to note is that at the very top of the full poll results, it says that this poll sampled 873 Democrats and 650 Republicans. Not quite an even amount.

I'm not a statistician by any means, but this is a good example of why you can't put all your eggs in the presidential poll basket.

2 comments:

Mommy In Pink said...

I never go by those polls...they are incredibly biased and largely liberal! Even though McCain is trailing in the polls, I think there are a heck of a lot more people out there that are for him than against him.

Anonymous said...

The disparity between the number of Democrats and Republicans surveyed probably reflects the total number of registered Democrats versus Republicans. If Dems outnumber Repubs by a 6 to 4 margin, then the survey will be targeted accodingly.

Karen

 

blogger templates | Make Money Online