Saturday, April 26, 2014

Gay Marriage, Bias and Poll Results

The Reverend Dog Gone is at it again.  She is trying to explain something to the religious right.

The only thing I found interesting in the little bit I read was this.  " A recent poll by KSTP,  arguably of a too few people to be very representative shows a majority of Minnesotans are pro-same-sex marriage, with a significant minority of those polled opposing."

Whenever I read a poll interpretation by a far right or left individual, I love to see how they are spinning it.  Now I am not sure where Dog went to school, but I am now concerned that math and statistics were not adequately taught.

Here is the actual text from the link.

"Last year the State of Minnesota legalized same-sex marriage. Do you approve or disapprove of that change in state law? Asked of 600 Minnesotans. Of the group 543 are registered voters.
47%     Approve
45%    Disapprove
7%      Not Sure
Margin of Sampling Error for this question = ± 4.3%"

Now I was taught that nothing can be determined when the numbers are 47%, 45% and +/-4. That is because the first number actually lies somewhere between 43% and 51%, and the second number actually lies somewhere between 41% and 49%.  Therefore the actual number of the disapproves could be smaller, bigger or the same as the approves.

Then you throw in the not sure answers and all bets are off.  So do you think this an honest mistake, a wishful thinking blind spot or something else?  Thoughts?

7 comments:

Anonymous said...

I tend to dismiss KSTP polls for the same reason I dismiss Strib polls; they are purchased news from highly partisan news organizations.

--Hiram

Anonymous said...

"Margin of Sampling Error for this question = ± 4.3%"

How do they know what the error is? Do they have some sort of actual number variation from which constitutes an error?

--Hiram

John said...

Roper Polling Fundamentals

My guess is that these polls are more biased by "who is willing to take the poll" than who is conducting it. Though how the questions are worded is very important.

This poll seems about right based on the amendment results.

Anonymous said...


My guess is that these polls are more biased by "who is willing to take the poll" than who is conducting it

I disagree. Liberal polling companies are hired by liberal news organizations like the Star Tribune; conservative polling companies are hired by conservative news organizations like KSTP. In each case, the news organization is purchasing the poll results it wants.

--Hiram

John said...

Interesting...

Now let's focus on just this survey result. Matts Maps

Remember that the 51.2% did not say they wanted gay marriage legalized. They simply said they did not want it made illegal via an amendment.
47.5 Yes
51.2 No
1.3% Blank

Do you disagree that gay marriage is a toss up?

Anonymous said...

Do you disagree that gay marriage is a toss up?

I have no real idea. It's certainly the case that the KSTP poll indicates a marked shift toward an increased popularity for gay marriage in a year and a half. But I really have no idea of the significance of those numbers. I think we are now passed the naive era of polling analysis when we view as significant isolated polling numbers from semi anonymous polling companies.

--Hiram

John said...

Doesn't look very anonymous to me...
Who is SurveyUSA