I believe you made a pretty big error in arriving at a proposed pregnancy rate of 2-3 per year. The pool of 37 women doesn't have 37 women in it in any given year. In any given year it probably has less than 10. Also, I think that extending the age limit up to 30 is pretty generous to your argument. I'm kind of doubting a 30 y.o.'s first concern is their parent's political fortunes.

You also seem to imply that all abortions are kept secret and could never come to light, which seems a strange argument in light of our very nosy political culture.

In these kinds of demographic exercises, the proper unit of analysis for these kinds of rates is not persons, but person-years. To make it easier, let's do it as person-election-cycles. Which, including the current prez/VP children, brings the number to 41, not 37.

However, you made a more serious error than this, as the above poster notes. In stating that your exercise gives you a pregnancy count of "two to three pregnancies per year," you are implicitly treating all of these women as though they were exposed to the risk of the event in each of the years of observation (years divisible by four, 1964 to present). But the event we're interested in here is unwanted pregnancies during a presidential political campaign. Assuming the time gap between presidential candidates (for whom the process lasts 1-2 years) and VP candidates (for whom the process may last only 2-3 months) washes out, this leaves us with two to three pregnancies among the group total, not per year.

Of course, that result assumes that these 18-29 year olds are likely to be similar to upper income 18-29 year olds generally in their unwanted pregnancy outcomes, but there's good reason to suspect otherwise. Unintended pregnancies are undesirable events by definition, but rare is the time when a child's pregnancy could ruin or threaten a parent's career. This added pressure and consequentiality of such an occurrence likely gives these children additional incentive to consistently use protection.

Your basic point may still apply, but the execution was disturbingly sloppy. Wikipedia could readily have taught you to do this properly.

There wasn't space to get into this level of methodology. Your 41 is not correct. The count starts when LBJ/Humphrey/Goldwater's daughters reach 17 and goes to the present, which is longer than 41 -- and each daughter runs multiple years in the sample (not 1 year, as you seem to be counting), so they overlap.

I never said the total number of sample years was 37 x (2008 minus 1964). It isn't, for the reasons you've mentioned.

I don't agree with you that the pregnancy has to be during the campaign cycle. It can screw up mom or dad's candidacy just as much if it happens a couple of years before.

Dear Mr Saletan, the first problem is that your sample is not statistically valid because it is not extracted from a random selection of the population under study.

But even if it were so, the issue as you state is irrelevant. The pregnancy of a teenager whose mother is in position of power to force her views on the rest of us and who advocates not only abstinence but forced pregnancy for victims of incest and rape, is indeed an indication that her mother's position is wrong and, even though it might not reflect on her parents, (after all who can control a teenager!), it does reflect on the harmful failings of her policies.

Actually, your fundamental methodological error was the assumption that this group of 34 women was representative of the broader sample from which the statistics were drawn. There is simply no basis for drawing this conclusion. The sample size is too small to be validated mathematically, and the facts which are known about these women clearly suggest that they are in fact atypical. It is not merely that they have higher education levels, higher incomes and better access to heath care than the general population, as you noted, but that they are the children of prominent politicians who are used to being in the public eye, and who are probably cautioned on a regular basis about the risks of media exposure. Most analysts would expect the unintended pregnancy rate for such a group to be much lower than that for the population generally.

Your fundamental logical error was the inference that because no pregnancies were reported among this group, the only possible explanation was that at least one of them must have had an abortion. That conclusion simply isn't warranted, first because your underlying (but unstated) premise that the individual women had no control over whether they became pregnant is fallacious, and second because random chance among such a small sample is as likely an explanation as your inference.

Your fundamental political error was the assumption that it would have been just as scandalous, and just as damaging, for one of the other women you named to have experienced an unwanted pregnancy as it is here. That conclusion is false for several reasons, including the fact that most of the other candidates were not strong advocates of abstinence-only education and most of the other women you named were adults, not minors.

This article was not only deeply flawed, it borders on slander.

Media exposure is a deterrent to lots of things. But do you really think it overpowers the human sex drive? (Look at Edwards, etc.) Or peer pressure? Or love?

Also, don't underestimate the girls whose parents don't happen to be politicians. It's not like they generally think precautions are less important just because mom or dad isn't running for office. But the unintended pregnancy rates show that some of them (and, yes, their boyfriends) still fail.

It's not logically impossible that none of these women had an abortion. It's just extremely implausible. And not because having done so makes them awful -- it just makes them and their boyfriends fallible, like the rest of us.

And no, it's not slander to say one of those 37 women probably had an abortion, any more than it's slander to say that at least one daughter of a ticketmate is probably gay, which now turns out to be demonstrably true. The statement probably is not true of any individual in particular, but probably is true of someone in the sample. That's just a paradox of statistics.

The "paradox of statistics" argument is valid only if the statistical analysis you employed is itself valid. Yours was not, for the reasons stated by multiple posters. Setting aside all the other issues, (btw, you don't seem to be garnering a lot of sympathy or support for this article), the purely statistical question is this: Given a sample size of only 34 women, with no knowledge of their individual fertility, sexual habits, or contraceptive choices, can one utilize national averages to conclude that it is more likely than not that one of them has had an abortion merely because none of them has borne a child. I think the answer to that question is obvious, and that you are wrong, but perhaps there is a statistician or logician out there who could enlighten me.

I am not necessarily inimical to your argument, but I don't think the statistics are particularly good. Of course there is a non-zero chance that some person in the group had an unwanted pregnancy, but to try to arrive at a convincing statistical argument, one is temped to manipulate one's assumptions until a satisfactorily high 'chance' is reached. If we narrow the age limit on the pool to just teen-age years, for instance, the percentage chance is probably unsatisfactorily small to make this argument.

squiffed:

If we narrow the age limit on the pool to just teen-age years, for instance, the percentage chance is probably unsatisfactorily small to make this argument.

That's an interesting idea. It might work out mathematically as you suggest -- but it also narrows the nature of a non-hypocritical critique of Palin.

I really wish readers would suspend the normal political urge for binary answers. Reality is way too complex for that. The answer may not be that criticism of Palin is or isn't hypocritical. It may be that understanding what would be hypocritical helps us think through exactly what we say about her.

Texvol:

(btw, you don't seem to be garnering a lot of sympathy or support for this article)

I hear you. I just don't think garnering support is my job. The political echo chambers of the Internet make people feel good about what they already believe -- they don't help us examine our assumptions or learn. My job is to make people think.

Your point would seem to be, "Let (s)he who is without sin cast the first stone," then demonstrating that the odds are "good" that "sin" has occurred, but again you start with the end in mind, and therefore your methods and assumptions are susceptible to confirmation bias... I don't necessarily disagree with your broader point, but if you employ quantitative methods to prop up an argument, you should expect that some will quarrel with those methods.

Making people think is an admirable goal, and I never meant to suggest otherwise. But making them think about whether one or more real women - women whom you actually name! - has had a secret abortion is of dubious value, particularly when the methodology you employ to create that inference is itself highly suspect. That strikes me more as an effort at sensationalism than provocation.

Assuming (somewhat generously, I think) that all you really meant to say was that Sarah Palin's daughter MAY not have been the first one to experience an unwanted pregnancy, and therefore we should be respectful in our reaction, you have still overlooked two important points : (1) the distinction between teen pregnancy and adult pregnancy, and (2) the irony of the situation in light of Sarah Palin's vocal - and unusually extreme - public policy positions condemning not merely abortion, not merely sex education in school, but most forms of contraception as well. It is entirely fair to point out that if Palin's policies were enacted on a national basis, we would have a whole lot more teen pregnancies in this country than we have now, and the confirmation of that fact is in her own house. No one could have said the same about John Kerry, or even Ronald Reagan.

This article was not only deeply flawed, it borders on slander.

Actually, it would be libel.

Actually I think there is a more serious error or at least a lack of explanation.

You say the following:

"Only 3 percent to 4 percent of women in the higher income bracket had unintended pregnancies in the two sampled years"

Even without looking at the source, it is logical to assume that that 3-4 percent is a percentage of the pregnancies. So that means 0.04 * 0.12 = 0.0048

(Otherwise it would mean 30-40% of all pregnancies being unintended which seems high)

I went further I assume youget the data from table B called "Total pregnancy rate, total fertility rate, percentage of births that were wanted at the time of conception, and wanted total fertility rate, by race and Hispanic origin: United States, 1996 and 2003"

Where you also go wrong is the definition Total Pregnancy Rate. The article explains:

"Total pregnancy rate
The total pregnancy rate and total abortion rate are estimated using the same procedure as for the total fertility rate. The pregnancy (or abortion) rates at each age are summed (and multiplied by 5 and divided by 1,000). The result is the number of pregnancies (or abortions) a woman would have in her lifetime if she experienced the age-specific pregnancy (or abortion) rates in that year."

So as I read this the "3-4%" you claim is in a womans lifetime! And not per year....