Could Survey USA be any less accurate

Posted by Jeff

Survey USA Polling Data from KSTP Poll

Wow, just when you thought SurveyUSA couldn't provide "less accurate" data they prove they can.


 

I am going to assume it's weighted to give 50% Men and 50% woman. The odds of that happening are pretty unlikely; however if they weighted it why not go with 52% women and 48% men to match what normally turns out to vote. Oh yeah since women favor Democrats over Republicans...

35% republican okay a little high but with such a small sample size we are talking a couple of respondents. 35+33+39=98 where is the other 2% you can't lose that in rounding.

Okay this one gets me 35% Republican which is a little high and 27% support of Tea Party. Those numbers need to be closer together. Again a small sample size where six people can change the results but....

86% of those surveyed vote in the Midterm elections, since this is a poll of likely voters for a Midterm election, does that seem correct to you?

53% of likely voters are college grads according to this survey. Considering Minnesota only has 28% college grads does that seem correct to you? Maybe the question was misrepresented because 55% of Minnesotans have some post secondary education.

The income question is another that is interesting, Minnesota has a Median income of $52K, so for only 32% to be more than $50K is an issue. That isn't one or two people in a sampling error that flat out wrong.

Here is my favorite; 94% of respondents are white, while Minnesota is mostly white it still has a 15% minority population. They simply cannot all be not-likely voters.

I won't get into the region of the State issue since in this survey it is either over represented or under represented depending on your definition of what is the Twin Cities Region.


 

Clearly this survey was directed at potential respondents who have a certain ideological view, can you guess who?

 

0 comments: