Curt Schilling for Bush
During a Thursday morning interview on ABC’s “Good Morning America,” Boston Red Sox pitching star Curt Schilling announced to everyone, “Tell everybody to vote, and vote Bush next week” (Inside Politics, Friday). Many were excited to hear this beloved pitcher endorse President Bush.
Later in the afternoon, it was reported the star would be traveling with the president in New Hampshire on Friday to two ralliesinManchesterand Portsmouth.
So, what happened between Thursday evening and Friday morning? After committing to appear with Mr. Bush, suddenly the Red Sox star announced he was not cleared to travel for medical reasons. Then came a statement from him that “speaking as I did the other day was wrong … it’s not my place, nor the time for me to offer up my political opinions unsolicited” (“President also campaigns with September 11 families,” Page 1, Saturday).
Worse, it seems the Red Sox are institutionally pro-Kerry. Team owners John Henry and Tom Werner and general manager Theo Epstein campaigned with Mr. Kerry in Manchester on Sunday, leading some of us to speculate whether the star pitcher’s political opinions were squelched.
Curt, many of your fans were very disappointed.
Not the picture of health
Mark Steyn’s “Election ennui” (Commentary, Oct. 25), in describing why he is “sick of this [U.S.] election,” is doubly ironic. The quality of his arguments is enough to make many readers “sick.” Indeed, the article’s comments about Canadian health care reveal the many myths that regularly appear in North American papers on both sides of the border.
Whatever the true state of Canadian health care, it is not revealed by a single type of bacterial infection (Clostridium difficile infection) in Quebec hospitals. This point neither supports nor criticizes the Canadian health system overall; it is a mark of Mr. Steyn’s faulty reasoning that he believes an instance settles the truth of a generalization.
For those who study North American health care, the picture is far more complicated. Quebec has indeed suffered sharp restraints on its health budget in the past decade and hospitals have been reorganized to such an extent that staffs have become strained.
But few in Canada would replace their rationing methods — which do not permit open queue-jumping with funds — to America’s system of incomplete insurance coverage and rationing by ability and willingness to pay. World-weary journalists of Mr. Steyn’s disposition are hardly guides to this subject.
University of Montreal
The cost of truth and justice
I found David Davenport and Gordon Lloyd’s Sunday Commentary article “Leave no vote behind” interesting. The last sentence reads, “Otherwise, we face the sad specter of this election being decided by lawyers in courtrooms, not by citizens in voting booths.”
I would like to point out that the lawyers will be making sure that each citizen’s bona fide vote will not be suppressed. Every vote should count and if it takes lawyers to make that happen, then so be it.
True, it might slow down the process of learning who actually wins this election in the timely manner we have been used to, but that is a small price to pay for truth and justice. Republicans and Democrats have no reason to fear this process if they have nothing to hide. Bring it on.
Don’t trust the numbers
Election Day is here and results from the latest polls seem to be occupying the public more and more (“It’s the turnout, stupid,” Page 1, Sunday). It is natural that attention turns to quantitative attempts to understand the voting landscape and predict the winner.
Polling companies are happy to oblige, conducting careful surveys that produce comforting quantitative information on a candidate’s standing. In addition, polls often report a sampling margin of error — usually 3 percent.
This 3 percent figure is quite misleading, for it suggests that the polling process can be analyzed rigorously, with the result being a small measurable amount of uncertainty. In fact this is completely wrong.
The sampling margin of error is not the only uncertainty in the polling process; it is simply the only uncertainty that is quantifiable.
Polls are sometimes careful to say that the 3 percent represents just one source of uncertainty, but what few people understand is that other sources of uncertainty, though difficult or impossible to quantify, are probably much more significant.
Even the most careful poll in a close election indicates only something that everyone already knew: That the election will probably be close.
The concept of polling as a way of predicting outcomes is straightforward. Imagine that all of the millions of ballots cast in this year’s presidential election are placed in a giant container, and are stirred around until they are thoroughly mixed. Then 1,000 ballots are blindly pulled out and tallied.
The percentage split of Bush and Kerry votes for these 1,000 ballots is quite likely to resemble the percentage split for all the votes. It can be mathematically computed that at least 95 percent of the time, the split of the 1,000 ballots will be within 3 percent of the actual split. This 3 percent is called the sampling margin of error.
In practice, however, no such giant container exists and it is impossible to find a truly random sample of 1,000 voters. The pollsters attempt to solve this problem as follows:
First, a collection of phone numbers are called and information is gathered from those households that are reached. Second, the raw data is normalized so the results are representative of the entire population, relative to factors such as gender, race and location of residence. Finally, polls re-weight according to a likely voter model.
Different polls perform each of these steps differently, but with the same goal: to simulate a representative sample of the voting population. How well is this goal accomplished? No one really knows and current conditions are adding further uncertainty to the process.
Specifically, the 8 percent of households with either no phone or only cell phones (up from about 4 percent in 2000) are not sampled.
In addition, response rates have dropped as more people than ever before are unreachable, too busy or unwilling to be polled. Combine these facts with the unprecedented number of newly registered voters and the fierce partisanship of this election, and it is clear that the polling models, no matter how well thought out, will contain a significant amount of uncertainty.
However, this uncertainty is ignored when polls are reported, and as a result the general public is falsely given the impression that this uncertainty is negligible.
For these reasons, it is misleading and irresponsible for the media to represent polling results as being accurate to within 3 percent. This is especially dangerous when so much of the news coverage is dominated by reports of minute fluctuations in the polls, and news of battleground states shifting ever so slightly from a light Kerry blue to the faintest shade of Bush red.
The polls tell that this election is likely to be a tight one, but nothing more. So vote today, the only poll that really matters anyway.
Assistant professor of mathematics
City College of New York
Professor of mathematics