The numbers on Quinnipiac University poll, discussed in my previous post, do not necessarily hold up under closer scrutiny.
1. Numbers that just don’t make sense
A. Litchfield County Numbers
Litchfield County, the smallest of Connecticut’s counties in terms of population density, was the only Connecticut county George W. Bush won in 2004. In this poll, Dodd is beating Simmons by 11%, and trailing McMahon by 17%. As evidenced by the head-to-head Republican primary numbers in this poll, Simmons and McMahon are drawing from the same group of Republican voters for their support, so why the huge swing between Dodd/Simmons and Dodd/McMahon?
In addition, the numbers continue to vary greatly in other Dodd/Republican head-to-heads. For example, Dodd is up 4% in Litchfield on Caligiuri (whose home base is over the border in Waterbury) but down 15% to Schiff and down 29% to Foley. So Schiff does 26% better than Simmons, McMahon does 28% than Simmons and Foley does 40% better than Simmons.
B. New Haven Numbers
The poll has Simmons beating Dodd in New Haven, which is hard to believe considering the overwhelming support Democrats enjoy in the Elm City, as well as in the 3rd Congressional district, held by Representative Rosa DeLauro.
2. Poorly Designed Poll
A. Registered Voters vs. Likely Voters
The Quinnipiac poll surveyed 1,236 registered voters. The poll does not screen for likely voters, which is obviously a more accurate method to gauge voters’ opinions.
B. Democrats, Republicans and Independents
Those polled were 474 Democrats; 332 Republicans and, presumably, 430 independents.In Connecticut, Republicans make up only about 20% of the electorate, so the number of Republicans polled to accurately reflect the break-down of the voters in the state should only have been about 247.
To be proportionate to the electorate, 519 Independents and 457 Democrats should also have been polled.
3. In addition, younger, more liberal voters who tend to rely on cell phones rather than landlines, are often underrepresented in polls like this.
