Taking a Stand: Sherrill poll was pretty accurate
Like many of you, the spread in the county executive race came as a huge surprise. Even the candidates admitted they were taken aback by the one-sided nature of the outcome.
Almost anyone I talked to leading up to the election said they believed the margin of victory between Kenneth Rogers and Carol Hamblen would be razor thin. After all, the race was decided by 20 votes four years ago, so there was a history of close finishes between these two. Even the poll finished days before the election showed the candidates neck and neck.
Long story short, Rogers ended up carrying all but three precincts and polishing off his rival by more than 1,700 votes. Nothing close to speak of in the end.
As far as the poll went, the county executive’s race blew it out of the water. The poll, which I finished the last day of early voting, showed them even.
While I could have understood this irregularity had the remainder of the poll been off, the rest of the poll was pretty well very close. In fact, it predicted the exact percentage of the trustee’s race which was won by Joy Slatton with 59 percent over Emily Panter with 41 percent. It was also very close in the Register of Deeds race where Terry Smith defeated Susan Curtis by a margin of 76 percent to 24 percent. The poll predicted a margin of 79 to 21 percent, well within the five-percent margin of error.
Even the circuit court judge’s race was very close. Stanley had 43 percent of the vote where the poll predicted 42 percent. The poll also correctly predicted the order of the other candidates in their finish. The only place where it messed up was putting Robert Newman too close to the eventual winner. The poll had him with 39 percent by he finished with 31 percent. Looking back at it, there are two reason for this.
First, the final three candidates each exceeded the poll’s predictions by slight margins. However, seeing they all exceeded slightly, the end result was a loss of votes for Newman. In addition, there was a sample I placed in the poll which I debated doing. I conduct the poll by adding bits of certain demographics. In this case, I believe the demographic for affluent males was too high. Had I left it out, my calculations show Newman would have had exactly 31 percent in the poll. Live and learn.
The sheriff’s race was also pretty close. The poll predicted Jackie Matheny with 81 percent. He finished with 74 percent. I believe it was the wording of the poll which took it off by just a couple of points. On the poll, we listed Mason Black’s name; however, he was a write-in candidate. In hindsight, I should have made people write-in his name, just as they had to do when they actually voted in the election. The fact his name was on there, made it easier for people to vote for him in the poll, thereby making his numbers higher and throwing off the votes of the other challengers. It’s one thing to say you’re voting for somebody. It’s another to actually use the write-in feature on the machine. People tend to not want to go to a lot of trouble when they are actually voting.
In all, the poll was once again pretty accurate, aside from the county executive’s race. Even in it, the poll missed Bob Warren’s vote total by just one percent, so it was pretty accurate there. Why was the executive’s race off on the top two finishers? I haven’t a clue. Remember, the people who voted for the other races in the poll were the same ones who voted in the executive’s race.
Once again, it goes back to the old adage that polls are only good for finishing. Sure, in most cases, the polls we took in May and July were accurate, but the executive’s race shows us we shouldn’t put a lot of stock in polls since the chance for error is always there.
(Duane Sherrill is a writer for the Standard. He can be reached at 473-2191 or by e-mail at [email protected].)