The Worthlessness of Political Opinion polls: A Case In Point

One of the most common tools for political research, and I use the word research very generously, is the opinion poll. Political candidates use polls to see how the voters feel about them, especially during or leading up to an election campaign. Political pundits, especially those in the media, use them to, in their eyes, substantiate their opinions, usually formed prior to the polls.

The problem with polls is that they are often misleading and generally worthless. Here’s why. Polls tend to use what they call samples from the population group, say all US voters, whose opinions the poll is supposed to represent. But these statistical samples are quite often fallacious and ridiculously too low to really represent the population they claim they represent. A classic example of this is a poll about who the voters of New York State feel should be appointed to replace Hillary  Clinton when she officially resigns her US Senate seat to become Secretary of State in the Obama administration.

The Quinnipiac University Poll, a private polling service that focuses on the Northeast, conducted this poll at least twice and compares the results of those two surveys. My first issue with their statistical sample is whether the two surveys, about three weeks apart, surveyed the exact same individuals or whether each survey asked the questions of two different, probably randomly generated, lists of sample New York State voters. I do not care how accurate they claim their statistical samples are, unless they surveyed the same individuals in both surveys, their claim of a statistical error factor of +/- 2.4% is pure fiction. Only by measuring the same sample in a study over time can you really claim to have an accurate picture of changes in that sample. Anything else is scientifically invalid.

My second and main complaint about the statistical sample of this survey is its very small size. They only surveyed 1664 people in the entire state of New York. The most recent figure I could find for New York State is 8,624,000 registered voters. That means their statistical sample was .02% of the total number of voters in the state. Answer one question for me, please: what is statiscally significant about 1 out of every 5000 voters? Not much, if you ask me.

I took a statistics course in college. Granted statistics has gotten more sophisticated in 25 years, but, if I remember correctly, a good statistical sample, especially of a large, very diverse populatiom like the voter of New York, should be at least one per cent or about 86,240. Had the people at The Quinnipiac University Poll gotten the same results with a sample of that size, I would take their results more seriously. I wonder who is paying for these worthless polls. Hopefully, not our tax dollars.

As I said above, polls often provide article material for the media, and bloggers. The problem is that these journalists tend to select only those portions of a survey that support whatever opinion they may want to foist on unsuspecting readers as scientific fact or the gospel truth.

One that I want to comment on as an example of what I said in the paragraph above is Real Clear Politics, a conserative blog and web site:

Founded in 2000 by John McIntyre and Tom Bevan, Chicago-based RealClearPolitics.com (RCP) has become one of America’s premier independent political web sites. Updated every morning and throughout the day, RCP culls and publishes the best commentary, news, polling data, and links to important resources from all points of the political compass and covering all the important issues of the day. RealClearPolitics has become a trusted filter for anyone interested in politics.

RCP’s political commentary, election analysis and polling averages have been featured in national media outlets including The New York Times, FOX News, CNN, The Economist, Investor’s Business Daily, The Chicago Sun-Times and many, many more.
About RealClearPolitics

In their post about the Caroline Kennedy survey, they focus only on the few questions about how Caroline Kennedy is losing the approval of New York voters, which the survey and the authors at RCP attribute to the poor beginning Ms. Kennedy had at her first live interview as a player in the political arena. Ms. Kennedy’s experience prior to the Obama campaign has been in museum and charitable foundation work. However, she does come from a very politically active and astute family, so I suspect she is quite knowledgeable about politics, After all, it was Ms. Kennedy who persuaded her uncle, Senator Edward Kennedy, to throw his very influential support to Mr. Obama. In my opinion it was the support of the Kennedys that sealed the nomination for Obama. And it is her role in that campaign that has all but guaranteed she will be the new Senator from New York.

What I found amusing about the RCP post was that, in the survey was a question about the media reaction to Ms. Kennedy’s somewhat nervous start to that interview in which her language was less than erudite, which the  voters of New York overwhelmingly said they expected the negative reaction of the media piranhas. But, TCP makes no mention of that question, Instead, they join the rest of the piranhas and imply that her poor preformance makes her unqualified to be a US Senator. All I can say about that is that if poor linquistic skills makes one unqualified for elected office, George Bush should never have been allowed to run for office.

I think Ms. Kemmedy’s poor performance was the result of a case of nerves. After all, she has led a politically sheltered life, probably since the death of her father, John Kennedy, in 1963. But given her academic credentials, an AB from Radcliffe College at Harvard University and a JD from Columbia University, graduating in the top 10 percent of her class there. Neither Harvard nor Columbia graduate people who do not have a high proficiency in the English language. I would venture to say that, normally, she is quite erudite. But she is also human and that means she is entitled to be nervous in what very well may have been a quite new situation for her. And, unfortunately, it is the nature of piranhas to go into a feeding frenzy at the slightest sign of blood.

Do I think Caroline Kennedy is qualified to be a US Senator? Most definitely. Certainly at least as qualified, if not more so, than Hillary Clinton was. And, as the survey points out, the vast majority of the survey respondents think Hillary did quite a respectable job of being a Senator. I think 51 years of being a member of the Kennedy clan should count for at least as much as Hillary’s eight years as Arkansas’ first lady and eight years as the First Lady of the US. Although she had not sought political office until now, I think her political savvy probably comes pretty close to that of her uncle.

In closing, I think Caroline Kennedy will make an excellent replacement for Hillary. If Caroline decides to follow the path of her uncle Ted and make the Senate her life-long home, she will end up a titan of that august body, and, who knows, perhaps in 2016, she may follow in her father’s footsteps and seek to become President. A Kennedy as the first woman president would not be so bad. We might even see a primary race between a female Kennedy and a female Clinton. No, not Hillary. Chelsea. Two daughters of former Presidents running for and one getting elected to the White Hous would rival this year’s election for historical prominence. And how’s the idea of one as President and the other as Vice-President sound?

Advertisements

One Response

  1. I wish I could say that it’s amazing how ignorant you are of basic polling practices, but sadly I’ve come to realize that most people share your ignorance. I suggest you either get your money back for the basic statistics course you took all those years ago, or otherwise find the closest university and enroll in their basic polling 101 course. It will take about 10 minutes into the first class for all of you objections and reservations about polling methodology to be refuted.

Leave a Reply

Please log in using one of these methods to post your comment:

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

%d bloggers like this: