« American History 101 | Main | Data, Reporting, and Blair »

Sample size

Lots of people, like Mickey Kaus, are going to town on the Blair story from the New York Times. Blair and the Times, these people are saying, prove that affirmative action is bad. Kaus writes that "fairly direct consequence of the Times's misguided race preference policy" (Kaus actually put this phrase in bold type).

John Leo at U.S. News and World Report writes:

"Everybody knows that this argument tends to trigger cries of 'Racism!' So let's stipulate: The overwhelming majority of plagiarism cases and journalistic scandals have been the work of whites.... But once you create preferences, you run the risk of increasing the number of screw-ups among the preferred group. Relaxing standards or pushing an unprepared candidate into a high-pressure job tends to increase the odds of trouble. All of us figure this out rather quickly when the preferred group is relatives of the boss or people who went to the boss's college. It's true of identity groups as well."

It's a cute little 'stipulation' on Leo's part, but it just doesn't wash. Blair isn't being accused of failing to meet standards that were relaxed for him. He has been accused of a crime. Leo knows this and plagarism is what he's been talking about until this paragraph. But then he equates standards with crimes. We have to throw away what Leo has written, unless he wants to rewrite his piece to "once you create preferences, you run the risk of increasing the number of plagarisers and criminals among the preferred group." Is that what he meant? If he did, you see why I don't buy his stipulation.

And, Leo is talking about trends. Trends. He uses terminology like "increasing" and "group" and "odds" -- that's trends talk.

In his talk about trends, how large is his sample? How many "unprepared candidate[s]" in this "preferred group" does his analyze? How much data does he have?

He, in fact, has a sample size of one. Blair. That's not trends data. He takes one datum point to paint with a wide wide brush. Not good argument technique -- not even in 8th grade.

So, Leo can't competently analyze data. (Or, he can, but choses, for rhetorical purposes, not to). He mix-and-matchs accusations (by equating 'low' standards with plagarism). Perhaps the New York Times isn't the only publication with editorial problems. U.S. News needs to edit its authors better, too.

TrackBack

TrackBack URL for this entry:
http://www.oobleck.com/mt3/mt-tb.cgi/711

Comments (3)

Richard:

Partha, nice try, but your attempt to change the subject from Blair to John Leo does not work. You may not like it, but an incident like this does call into question the whole concept of affirmative action. You are trying to say that Blair was a criminal (that's kind of strong language, isn't it?) because he committed plagiarism. However, that was only a small part of his actions. Most of the time he simply made up his news reports, as, for example, the comments that he attributed to the alleged DC sniper.

Leo's point was that Blair was in over his head and was not able to perform at the level expected of a NY Times reporter. (Of course, some of us would say that given the reporting by Johnny Apple on the war in Iraq, he was par for the course.) And that of course is the tragedy of affirmative action as practiced by the NT Times. The Times makes hiring a minority, not a good reporter who happens to be a minority, an end in itself. This policy calls into question the competence of all minorities on the Times, no matter how unfair that may be.

And, Leo is talking about trends. Trends. He uses terminology like "increasing" and "group" and "odds" -- that's trends talk.

Once again we can add a new word to the Partha dictionary. From the real dictionary: trend - (1) the general course, drift or tendency, (2) to tend to take a particular direction.

Odds have nothing to do with trend. Group has nothing to do with trend. I will agree with you that one incident does not make a trend. However, that is not what Leo was talking about. His point, which you do not have to agree with, is that when you give preference to the hiring of a minority, simply because the person is a minority, then you are more likely to get someone who is unqualified.

Partha Mazumdar:

> Leo's point was that Blair was in
> over his head and was not able
> to perform at the level expected
> of a NY Times reporter.

Do students cheat on exams because, even if they had studies, they would have failed or because they are cheaters?

I think it's because they are cheaters. It's not because they couldn't have done the work if they had tried.

> This policy calls into question
> the competence of all minorities
> on the Times, no matter how
> unfair that may be.

If we both agree that it's unfair, let's not call their competence into question. No one is forcing us.

> His point, which you do not have
> to agree with, is that when you
> give preference to the hiring of
> a minority, simply because the
> person is a minority, then you are
> more likely to get someone who is
> unqualified.

Leo gives his readers absolutely no reason to agree with him. He bases his entire column on one datum point. One can't talk about "more likely" when you only have one piece of data to point to.

So, yeah, you're right. I don't believe him.

About

This page contains a single entry from the blog posted on May 12, 2003 4:23 PM.

The previous post in this blog was American History 101.

The next post in this blog is Data, Reporting, and Blair.

Many more can be found on the main index page or by looking through the archives.

Powered by
Movable Type 3.31