Thursday, May 08, 2008

Who Do You Trust?

Here is a little report card on the polling organizations from Survey USA. It shows the average median error of polling organizations that conducted polls, within seven days of a presidential primary election. The lower the number, the more accurate the polling organization has been.

I do not particularly like the market being flooded with polls prior to elections, because I think they can persuade some voters from voting. If their candidate is way behind in a pre-election poll, they may just think their vote will not count and blow it off altogether. If the poll is skewed, if enough people see it and do this, it effects the election.

But, no one cares what I think. So, poll they do. At least this gives us an idea which polls are more likely to have the more accurate information.

Check it out, when you get a chance.


Addendum:

For what it's worth, here is Survey USA's average mean error report.


6 comments:

Obob said...

I tell my students polls are for propaganda and conversation.
You can use them to make your candidate look stronger and manipulate the issues. Or some mindless number to make yourself look smart at the expense of doing any real research.
Although the exit poll on Hillary was scary close to reality for once. 52/48

Greg said...

But, no one cares what I think.

People come here every day to read what you think!

So, I guess we should be reading polls from Suffolk University, Survey USA, Zogby and...Fox.

Mustang said...

"I think they can persuade some voters from voting.”

Absolutely. They also convince voters to pursue the (perceived) "more traveled path." By any definition, polls are no more than a cursory indication of . . . whatever. The samples are limited and the argument for extrapolation doesn't stand up. What should we expect from a poll of 100 respondents in downtown Boston? Does this reflect values and opinions of potential voters in any other state? Hardly.

Suppose we used that same polling data to form conclusions about the voters themselves and then publicize it. That would be a hoot.

I posted on a similar issue today. What is your view?

Semper Fi

LASunsett said...

//I tell my students polls are for propaganda and conversation.//

And for talking heads to figure out what the issue of the day is, so they can get their lies straight.

LASunsett said...

//So, I guess we should be reading polls from Suffolk University, Survey USA, Zogby and...Fox.//

What really was surprising to me was where Rasmussen placed. It is widely regarded as one of the more credible polling organizations.

People are constantly accusing FOX of bias and skewed polling, but they placed fairly well.

LASunsett said...

//Suppose we used that same polling data to form conclusions about the voters themselves and then publicize it. That would be a hoot.//

There are many directions and ways I could steer my response to this one, but this is a family blog. Suffice it to say that there are many people that do not have the common sense of an animal and I think polling would bear this out, only if the sample was truly honest.

;)