A number of posts have discussed problems with survey research. At Politico, Steven Shepard reports that public pollsters tended to underestimate GOP support in the 2014 midterm.
The results were another black eye for pollsters in what are already some tough times. Just five months after then-House Majority Leader Eric Cantor’s surprise ouster, it was another out-of-nowhere Virginia race that left political observers scratching their heads.
And as Americans become even harder to reach by phone – and emerging methodologies, such as Internet polling, remain unproven – the poor performance of pollsters this year casts serious doubt on the reliability of surveys during the 2016 presidential race.
...
Republicans have long claimed that public polls, usually conducted by randomly dialing phone numbers rather than only contacting voters with a history of turning out in midterm elections, include too many people who won’t ultimately cast a ballot – a group that tends to lean Democratic.
Those public surveys, they say, also weight, or peg, their demographic data to known Census parameters, ignoring historical trends of the midterm electorate – which is usually older and more white.
“I think the media polls were dramatically off because too many media pollsters use Census weights,” said Republican consultant Brad Todd, whose firm, OnMessage Inc., conducts polls and creates TV ads for GOP candidates. “In a midterm electorate, using the Census as a reference point would have the same value as using a grocery list as a reference point.”
