Why Polling Shouldn’t Be Reported as News

When I was taking my graduate statistics course it was a point of debate in the class on how you can, through manipulation, make statistics infer anything you want.  The point of the discussion was to undersatand how personal bias and agenda can potentially cloud the results of even the most rigid surveys or samples.

Today our news channels inundate us with poll after poll purporting to show how Americans feel about almost any topic you can think of.  Each poll is presented as indisputable evidence of whatever agenda the news organization wants to push.

For example, lets take the latest AP Poll - Racial Attitudes Topline (Oct 28), being pushed around the news world.  It is supposed to show how average American’s have become more, not less, racist since the election of President Obama.  The Washington Post felt compelled to defend the legitimacy of the poll in this article,  Washington Post -- How the Associated Press polls on racial attitudes were conducted, but in that defense I see no explanation on how academic bias was accounted for, or how they legitimatized the assumptions on human performance in measuring implied feelings from association.

So what kinds of Headlines come from this online survey? 

Washington Post – “AP Poll: A slight majority of American’s are now expressing negative views of blacks”

 The Guardian (UK) – “Racial prejudice in US worsened during Obama’s first term, study shows”

 NY Times – “The Price of a Black President”

The Hill (blog) – “Poll:  Racial attitudes towards blacks could cost Obama at the polls”  (I would recommend the Hill get a new Headline writer if this is the best they can do).

You know, I read the poll numbers, and there is a minor shift between 2010 and 2012 in approval ratings for the President, but then the economy still stinks and we’ve listened to two negative campaigns slamming each other for the past year, so shouldn’t we expect a shift in approval ratings?  As far as the race questions all three categories, White, Black and Hispanic shifted but I am not sure how in the limited sample size they had you could say with certainty it was beyond the margin of error found in any polling sample.

I wonder if the fact that any time someone is critical of the President some media type feels compelled to label that critic as a racist would have anything to do with these findings?

There is one thing I would take as a certainty.  Chris Matthews of MSNBC will have a field day with this poll.  I doubt he’ll read it, but I would expect he will cite it often, without any realization of his potential role in the shift, if in fact there is one.