Beware the snake oil salesmen peddling 2014 election surveys -- By Mark Tapscott, The Washington Examiner

It was utterly predictable that liberal activists and their traditional media allies would claim that Republicans will lose their House majority in 2014 as a result of the government shutdown.

It was just as predictable that their Exhibit A would be a severely flawed poll bearing the same resemblance to statistical validity as Mr. Twain's lightning bug had to lightning.

Remember Moveon.org?

Moveon.org commissioned the Public Policy Polling survey now being touted throughout the traditional media as proof the Republicans are committing political suicide by holding firm in the government shutdown battle.

As the Washington Examiner's Tim Mak reported Monday, the PPP survey of 24 Republican-held swing districts "shows that a generic Democrat would beat a Republican incumbent in 17 of them -- the very number of seats Democrats would need to wrest from Republicans to take control of the House."

Be advised, however, that the PPP poll is all but worthless, for many of the same reasons most such exercises in data manipulation are taken seriously only by ideologues. Consider just two of those reasons.

First, the PPP survey has two major flaws. It covered only 700 adults, so the sample size is less than desirable for measuring voter behavior. Far better is a survey sample of at least 1,000 individuals.

Ask the Foreign Legion:

More serious is that the respondents are randomly selected adults, not likely voters. As a result, a survey of 700 members of the French Foreign Legion would have been about as useful for predicting the outcome of the 2014 congressional contests.

The bigger issue with such surveys, however, is one former New York Times survey guru Nate Silver explained a few years ago, namely sample bias against Republicans, which is especially evident in polls based on adults rather than likely voters.

 (Click link below to read more)
READ MORE Sphere: Related Content