Earlier today, at Instapundit, Sarah Hoyt linked a post which showed how Quinnipiac weighted their latest Ohio poll to create a bigger Democratic advantage than the sample provided. “Its partisan split“, blogs Bryan Preston, “skews much too far to the left:”
So, I tallied the raw totals (unweighted frequency above) and came up with a sample size of 1,110. This means, the Republican total of 343 represents 30.9% of the sample, the Democrat 35.7%. Thanks to Quinnipiac’s ratings, suddenly presto-chango a D+4.7% sample becomes D+8 sample, pushing the race 3.2 points toward the president’s party.
It seems that the raw numbers only had Obama up by a point. Guess Quinnipiac just didn’t like that result.
Methinks the folks at Quinnipiac were trying to get the Democrat to 50–which is what their weighted survey showed.
ADDITIONAL THOUGHT: Why does Quinnipiac forecast a large Democratic advantage this fall than in 2008?
FROM THE COMMENTS: Kurt writes that, “Of course we all know why they did it–so as not to depress turnout!”
Can you imagine what will happen if Ohio or Pennsylvania are decided early for Romney? If that were to happen, then some of the states further west might change, too.