Yesterday, as I was looking over the news, I came across an article noting that Governor Mitt Romney now has a four point lead in the presidential race after Pew conducted a post debate survey. The poll is the first of many to be conducted by the Pew Research Company. Pew's survey showed that viewers of the first presidential debate "overwhelmingly believe Romney won the debate: 66 percent said the republican candidate performed better, compared with only 20 percent saying Obama performed better".
After reading over the article, I found the survey methods at the bottom of the page to be extremely interesting. One has to question where all of Pew's numbers come from. According to the article, Pew conducted live phone interviews from October 4th through 7th with 1,201 registered voters and 1,112 likely voters nationwide. The margin of error is 3.3 percentage points for registered voters and 3.4 percentage points for likely voters.
Phone surveys are quick and efficient, but leave many questions. Was the population Pew interviewed representative of the entire nation? There is little information on where the sample came from; perhaps the majority of respondents were from a republican-favoring state. What was the method of the phone survey? Did numbers come from a random-digit-dialing technique or phone lists? How did Pew verify that the interviewee was a registered voter? All of these questions come with a phone survey. Perhaps another type of survey, like an in person interview, would reduce the amount of questions that this survey has seemed to ignore.