With Election Day less than three weeks away, it seems that every day another poll is released and then over-analyzed by the media. This so-called “horse race” journalism, where talk about who’s up and who’s down predominates instead of a substantive discourse on the issues, is not especially constructive and is often predicated on faulty or even biased data.
Major polling outfits usually don’t conduct surveys of local races for State Senate or Assembly, so oftentimes individual campaigns do their own internal polling. Therefore, it was relatively big news a few weeks ago when Siena polled four competitive State Senate races, including the Frank Padavan-Tony Avella matchup here in Queens.
The Marist poll in September that showed Carl Paladino within six points of Andrew Cuomo momentarily turned the New York political world on its head and penetrated Cuomo’s aura of inevitability. However, this poll was an outlier, and other polls taken around the same time and more recently have Cuomo leading by double digits.
Polling is an art rather than a science. Conventional wisdom is that Rasmussen oversamples Republicans, while earlier this year the progressive blog Daily Kos filed suit against Research 2000 for supplying fraudulent data for polls they conducted.
The actual statistical reliability could also be hit or miss. The margin of error is usually between 2-5 percent, and increases for smaller subgroups, like women or seniors. And then there is the 5 percent probability that the true numbers are outside the margin of error completely.
The other problem, especially in a midterm election cycle, is assessing likely voters as opposed to registered voters. Of course, there is the natural human inclination to give the “right” answer when responding to a survey question, so when a pollster calls and asks whether you are likely to vote on November 2, most people will say yes, whether or not they really plan on voting.
Of course, pollsters have ways to figure out whether someone really is a likely voter, such as vote history and asking whether they know where their polling site is, but such methods could be hit or miss. This is one reason why polls, especially for the primaries this year, have not been particularly reliable indicators of election results.
So caveat reader, or viewer, when pundits and talking heads starting spouting off on poll results in the next few weeks
Daniel Egers is executive director of the Queens Republican Party and an advisor to the Assembly campaign of Vince Tabone.
