« 
: Tough questions... |
Main
| Dramatic risks... »
October 21, 2003
What the numbers say...
I've tried twice (a low figure) to interest reporters in the Mayer predictive model of primary campaigns and my use of it to suggest that reporters spend more time covering policy during the pre-primary period.
Silence.
I suspect the reason is the reporters think I'm crazy. After all, everyone knows that the primaries are competitive races based on the candidates' performances and strategies, and each race builds on the other leading to a winner. The model, however, says this isn't true.
Like any predictive model, this one could fail. But, considering its consistency (correct since 1980), I'm sticking with it. The question arises: Why is the model so consistent? I think we can find the answer in the Central Limit theorem: "the average (or sum) of a large bunch of measurements follows a normal bell-shaped curve even if the individual measurements themselves do not" (re: John Allen Paulos). That's pretty much the same as saying: you can't see the forest for the trees.
The forest in this case is the national Gallup poll charting the preferences of likely voters. The trees are the individual state polls. Despite plenty of evidence to the contrary, many journalists continue to think that winning in Iowa, New Hampshire, or any other given state is necessary to win a nomination. They continue to believe that charting such wins and losses indicates momentum.
(What would it mean, in terms of journalistic practice, if these things turned out to be misleading or false?)
I think the Mayer model demonstrates that the Central Limit theorem works even in a complex political system. Voters in the aggregate confirm the national polls with their votes because the national polls accurately represent the normal distribution of voter preference. The individual state polls, then, constitute too much detail, i.e. detail that is not predictive of the larger primary system.
The tricky part this year: statistically, there is no front runner yet. For the model to work, someone needs to emerge, i.e. take a lead that exceeds the margin of error.
If this doesn't happen, we will technically have (in a non-pejorative sense) chaos.
UPDATE (22 Oct. 11:11 a.m.): Jay Manifold responds.
Posted by acline at October 21, 2003 10:40 AM | | Spotlight