Nader's reach won't cut Kerry

Here's the unedited version of a letter I had printed in the Washington Times back in March. The version they printed is here. (If I recall, the major difference was they cut out the part where I asked them to be accountable.) The situation may certainly have changed since then, but I think the overall sloppy statistical analyses are prevalent in papers everywhere. The bigger picture, as it pertains to the presidential race, is this confusion of strategy for fact -- it's usually just an accepted truth that Nader's success is universally bad for Kerry and universally good for Bush, a corollary to the "Nader made Gore lose" train of thought.

To the editor:

I wish to comment on a bit of sloppy analysis that slid its way into Steve Miller's article "Nader campaign gives Bush a boost in polls" (The Washington Times, March 17, 2004). I'm no statistician – and clearly, neither is Mr. Miller – but these results offer few clues as to how many votes Mr. Nader seems to be pulling from any candidate (assuming for the moment that the votes would otherwise be "owed" to some candidate). Mr. Miller compares results in which Mr. Nader was included in the poll questions to results in which he was not, showing a dip in Mr. Kerry's numbers of 5% (the margin of error was 3%).

But his extrapolation includes no discussion of the unpledged voters responding to either question. According to the full poll results, when the question posed mentions only Mr. Kerry and Mr. Bush, 1% vote for "Other," 1% said they "Won't vote," and 2% said "Depends." When Mr. Nader is included, the total from those categories was 1%. In other words, Mr. Nader picked up 3% of his 7% total from people uncommitted to either major-party candidate. Given the margin of error, one can only definitively confirm that Mr. Nader drew 1% of voters from Kerry – hardly "cutting a wide swath," to use Mr. Miller's words.

Also missing from Mr. Miller's analysis were the total number of nonvoters polled – pointing out things like 18% of the 1,206 adults polled were not registered voters, or that 29% of everyone polled did not vote in 2000, might have given some perspective on the helpfulness and reliability of questioning this particular group of people. Furthermore, no questions were directly asked of those being polled about Mr. Nader's effect on their votes, which would have been the only sure way to gauge the bearing on Mr. Kerry's support.

Perhaps Mr. Miller did not have space to properly analyze the poll or its results. But to make unqualified statements about "a [continuing] pattern of Mr. Nader dipping into the well of votes Mr. Kerry is expected to secure" is intellectually lazy, bordering on dishonest. This is particularly so when the only evidence given to corroborate that statement is another poll showing Mr. Nader's support among members of his ethnicity in four unnamed states. Considering the headline and focus of this story on Mr. Nader's influence on the race, one would expect some more evidence – indeed, one would expect more of Mr. Miller and this newspaper.

Sincerely,
Mark Zipkin