The Poll Next Tuesday

Yesterday, WISH TV and Franklin College released the first independent poll of the mayoral election. It had something for everyone–Ballard is polling significantly less than 50%, typically a danger sign for an incumbent. On the other hand, he was ahead of Kennedy. The number of undecideds was huge this late in the race, and the early voters favored Kennedy by a wide margin, suggesting more enthusiasm for the Democrat.

Of course, when a poll has a 4.9% margin of error, any results should be viewed with some skepticism.

Actually, political polling has fallen on hard times. Last year, Brian Vargus–the Political Science Professor who has long been regarded as the local expert on political surveys–came to talk to my class. His message was that most political polling is worthless–that accurate, reliable surveys are prohibitively expensive, and campaigns and media outlets simply don’t do them anymore. That’s why you see such large margins of error.

A good poll will be representative of those actually likely to vote. That means including minority communities that are historically under-polled (Julia Carson routinely polled 15-20 points lower than her vote on election day). It means including younger voters who use cell phones exclusively (and accounting for the fact that they’re less likely to vote). In other words, a good poll requires both accurate sampling and┬áthe use of methodologically predictive algorithms.

According to this morning’s news, Vargus raised several questions about the WISH poll, based upon some internal inconsistencies. But even assuming it is reasonably accurate, what it tells us is what politicians have always known: what matters is the poll taken on election day, and that poll depends on who does the best job getting out their vote.

A recent study of 155 elections involving incumbents showed that voters who were undecided two weeks prior to the election broke 80%/20% for the challenger. Evidently, if the incumbent didn’t have them by then, he wasn’t going to get them. That’s good news for Kennedy–if her campaign gets those folks to the polls.

At the end of the day, turnout is the key.


  1. Actually margin of error is a possible statistical variation based on the size of the sample versus the overall population. I don’t 100% understand it. I do know that reporters misinterpret it when they say a 5% difference between the candidates is outside of a 4% margin of error. It is actually 4% on either side of the candidates’ numbers. So it has to be more than an 8% spread to be outside a 4% MOE.

    The media overly focuses on the MOE when they’re trying to examine polls beyonde the MOE. Much more important is such thing as the sample that is selected, that it is a good cross example of those who will be voting. Vargus points out a problem with cell phones. I would add to that caller ID and voice mail. People just don’t pick up the phone like they used to. Pollsters generally don’t have cell phone numbers, so what they do is poll more younger voters who have landlines to make up for the large percent of younger voters who have only cell phones. It’s really a tough adjustment for cell phones and for other incidents of modern technology.

    What is most signficant about the WISH poll is the 44% number Ballard received. Anything below 50% is considered to be big time trouble for the incumbent. Then you also have 21% undecided (or refused to answer) which is astonishingly high. Those people normally break sharply for the challenger.

    If an incumbent mayor told me he was at 44% in a poll with 21% undecided in what was essentially a two person race, I’d tell the Mayor to start getting his resume ready.

  2. I would argue that in a city that has a heavy democrat registration and Melina only having 33% that she is the one in trouble. She is not able to get her own base to support her.

    Don’t be surprised by a double digit loss by her next week.

  3. I agree with Paul on this one. Also, remember last year in Nevada right before the congressional election when Sharron Angle was polling ABOVE Harry Reid in EVERY POLL. In fact, on election day the poll showed her 5 points ahead while he was still in the 42% range. Then, there was a massive GOTV during the election and Harry Reid won 50%-45% over Angle. It goes to show how aggressive GOTV is and what is can do. I recently spoke with the Kennedy Campaign manager, and he told me that this year will be the biggest, most agressive, and most sophisticated GOTV effort this county has ever seen.

    Another example: Bart Peterson was at the 43% mark right before the election in 2007 and he LOST. Whenever an incumbent is below 50% THIS CLOSE to an election it is never a good sign.

    Paul, if I were speaking to the mayor I would be telling him to be prepared for the worst.

Comments are closed.