Sunday, April 26, 2009

Strib bias

Election night results for MN Senate race-

Independent Dean Barkley 15.15%
Republican Norm Coleman 42.98%
Democrat Al Franken 42.99%

Party preference results from new StarTribune poll-
Republicans 20%
Democrat 36%
Independent 37%

Headline-Most want Coleman to call it quits

The Strib poll called random adults, without asking if they had voted in the last election. Nonetheless, they oversampled independents by 20% compared to those that voted on election night. Given the voter turnout for Minnesota in 2008, at least 20% of these poll respondents likely did not vote for US Senate.

Results of a poll based on 1,042 interviews will vary by no more than 4.0 percentage points, plus or minus, from the overall population 95 times out of 100. Margins are larger for groups within the sample, such as Democrats and Republicans.

What does that last line mean? It means that at 1000 respondents, the actual results could be off by 4 points either way. Smaller sample groups mean a larger margin of error. But that doesn't stop the Strib from giving more headlines.
Although 57 percent of Republican poll respondents approve of Coleman's appeal to the state Supreme Court, the same portion of Republicans want him to quit should he lose there.
Since only 20% of poll respondents identified themselves as Republicans, the number of Republicans in the poll is about 200. The usual margin of error in a poll of 200 is at least 7 points. Which means that it could actually be 50% or 64% of Republicans that approve of Coleman's appeal.

Buried in the story is this interesting tidbit-
But two-thirds said the election system either needed improvement (46 percent) or was seriously flawed (21 percent). Just 30 percent said they thought it generally accurate and fair.
Given the fact that Republicans represented only 1 in 5 responses, that means a significant number of Democrat and Independent respondents think there are problems in River City.

No comments: