Did you see the recent report that the odds are now 51% that the Republicans will win the Senate this year? That number is down a bit from a the previous estimate. The truth, however, is that this sort of analysis is total garbage. Those making the estimate use polling, history and their own analysis of each state to come up with an estimate for each race and then combine those estimates to get an overall estimate. It sounds good, but it is really nonsense.
Think for the moment about the estimates whether or not it will rain tomorrow that you see in the weather reports. Is there really a meaningful difference between a 50% chance of rain and a 60% chance of rain tomorrow? I don't think so given the inherent variability of the weather. Now consider that there is a prediction of the likelihood of rain in seven weeks on election day in each of the 50 state capitals. Would you put much stock in a prediction that called for rain in a majority of those locations? I don't think so.
Then there are the problems with using the past to determine the future in politics. In the latest analysis, we are told that the likelihood of senator Udall being re-elected from Colorado is almost 100%. I am sure that the pundit who did the analysis looked at all the trends in the past that appeared in Colorado and added them to the poll results. But then along comes a poll like today's in USA Today. It finds that the Republican candidate is beating the Democrat in Colorado by 1%. That does not mean that the GOP will win, but it certainly ought to mean that it is not 99% likely that the Democrat will win.
Until there are big and consistent margins showing in the polling, it is hard to decide that one side or the other will win a race. Look at the Ohio governors race. Incumbent Republican John Kasich was ahead in the polls in his fight with Democrat Ed Fitzgerald even though Kasich had fallen in his approval numbers last year. Most pundits would have predicted a Kasich victory, but it was far from sure. Then, Fitzgerald was discovered in a parked car at 4 a.m. with a woman who was not his wife. The polls now give Kasich a 30% lead. We can probably call that one.
The truth is that less attention needs to be paid to the polls. Wouldn't it be better to discuss the positions taken by the candidates rather than their poll numbers? Somehow, however, the mainstream media cannot bring itself to cover the issues rather than the polls. I say we ignore it all.
Think for the moment about the estimates whether or not it will rain tomorrow that you see in the weather reports. Is there really a meaningful difference between a 50% chance of rain and a 60% chance of rain tomorrow? I don't think so given the inherent variability of the weather. Now consider that there is a prediction of the likelihood of rain in seven weeks on election day in each of the 50 state capitals. Would you put much stock in a prediction that called for rain in a majority of those locations? I don't think so.
Then there are the problems with using the past to determine the future in politics. In the latest analysis, we are told that the likelihood of senator Udall being re-elected from Colorado is almost 100%. I am sure that the pundit who did the analysis looked at all the trends in the past that appeared in Colorado and added them to the poll results. But then along comes a poll like today's in USA Today. It finds that the Republican candidate is beating the Democrat in Colorado by 1%. That does not mean that the GOP will win, but it certainly ought to mean that it is not 99% likely that the Democrat will win.
Until there are big and consistent margins showing in the polling, it is hard to decide that one side or the other will win a race. Look at the Ohio governors race. Incumbent Republican John Kasich was ahead in the polls in his fight with Democrat Ed Fitzgerald even though Kasich had fallen in his approval numbers last year. Most pundits would have predicted a Kasich victory, but it was far from sure. Then, Fitzgerald was discovered in a parked car at 4 a.m. with a woman who was not his wife. The polls now give Kasich a 30% lead. We can probably call that one.
The truth is that less attention needs to be paid to the polls. Wouldn't it be better to discuss the positions taken by the candidates rather than their poll numbers? Somehow, however, the mainstream media cannot bring itself to cover the issues rather than the polls. I say we ignore it all.
No comments:
Post a Comment