Comments on FiveThirtyEight: Politics Done Right: Pro-Life States Have Lower Abortion Rates

Looks like some very fishy "science going on here ...

2009-10-19T19:15:59.997-04:00

Looks like some very fishy "science going on here between 2000 and 2001:

http://www.data360.org/dsg.aspx?Data_Set_Group_Id=528

酒店經紀人, 菲梵酒店經紀, 酒店經紀, 禮服酒店上班, 酒店小姐兼職, 便服酒店經紀, ...

2009-10-10T08:09:01.399-04:00

酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店經紀,
酒店打工經紀,
制服酒店工作,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
酒店經紀,
菲
梵,

@Tom: Great post on the weaknesses of the data. ...

2009-10-09T04:44:15.969-04:00

@Tom: Great post on the weaknesses of the data. I too was taken aback at the beginning, wondering if all doctors were really required to report how many abortions they perform in all of those states (which would seem necessary for the data to be unbiased enough to trust).

Considering the strong social stigma still attached and the lack of safe and anonymous reporting methods I think the null hypothesis is that people still underreport abortions (and more so where--surprise!-- it's more stigmatized).

Any true analysis would have to take account of -percent of all (unwanted) pregnancies-, which outside of a police state coming around, seems unlikely, especially with the new availability of the morning-after pill prescription-free.

Theonlysaneone: Making abortion illegal would eff...

2009-09-16T22:59:12.738-04:00

Theonlysaneone:

Making abortion illegal would effectively reduce the number of abortion clinics to zero. In any case, the legal abortion rate spiked over the first few years after abortion was legalized. And if defunding abortions reduces their incidence, imagine how much more effective making it illegal would be.

theonlysaneone: This data, while interesting, does...

2009-07-14T13:42:33.171-04:00

theonlysaneone: This data, while interesting, does not seem to have any policy implications, except that pro-lifers should work on decreasing the number of abortion clinics if they want fewer abortions to be performed.

I think they're quite aware of that.

Not a very responsible post Mr. Silver. The data ...

2009-06-23T18:48:29.025-04:00

Not a very responsible post Mr. Silver. The data caveats seriously implicate your conclusions, check it:

http://feministsforchoice.com/?p=649

@theonlysaneone: "You seriously believe that any c...

2009-06-05T12:40:10.495-04:00

@theonlysaneone: "You seriously believe that any child would rather be dead than live in foster care?"

Generally, yes. Have you asked any?

I could recount three personal stories told to me by adults who were put into the system as infants and passed around in foster care before adoption at ages 5 (the youngest) to 12 (the oldest). All three consider their upbringing to be a black cloud over their lives. The two adopted at older ages said that they would have been better off not born, and that their upbringings were cruel. All three have notable psychological trouble dealing with careers and relationships.

Overpopulating the foster care system with, essentially, discarded children (because of inability to get an abortion, or instilled belief against it) is cruel to the children involved.

And I haven't even addressed the topic of pregnancies (teenage or otherwise) where the child is kept, but the financial burden of that choice ends up having dramatic negative repercussions. In situations where the family ends up falling [much further] below the poverty line (and the US's paltry government assistance can't pay the bills), where the parent(s) can't care properly for the child, where the parent(s) must depend on known-abusive family members to care for the child... those situations are not also cruel? (You can find plenty of cases like these in any inner city. Just drive on in to "those" neighborhoods you normally avoid, and talk to a few parents with young children. You might learn something.)

@Bob X: "these same people are out to forbid same-sex couples from adopting"

As part-time activist, I know this only too well.

I am pro-life so please don't take this the wrong ...

2009-06-04T14:18:45.700-04:00

I am pro-life so please don't take this the wrong way when I say that your analysis is flawed.

You have employed what is known as the "ecological fallacy". It is a common statistical fallacy in the social sciences. The wiki (http://en.wikipedia.org/wiki/Ecological_fallacy) does a fairly good job of explaining it, but the original paper explaining it (Robinson, W.S. (1950). "Ecological Correlations and the Behavior of Individuals". American Sociological Review 15: 351–357) is far more clear in my opinion.

The short version is this: when you plot "percentage of X" vs. "percentage of Y" you can't make any conclusions about individuals. Depending on how you group those individuals you will get different correlation coefficients (state, county, district, etc.). The correlation coefficients obtained from such plots are not meaningful in the context of testing hypotheses, as P-values for hypotheses such as "H0: m = 0" (for a proposed model Y = m*X + b) are erroneous. Drawing conclusions from such "correlations" is no more scientific than pure speculation.

There may in fact be something to your working theory, but your figures and Excel-calculated "correlations" do not substantiate it.

Ole Forsberg -- you said "The number of pregnancie...

2009-06-03T19:10:34.115-04:00

Ole Forsberg -- you said "The number of pregnancies equals the number of abortions plus the number of live births less the number of failed pregnancies (unless I am forgetting an alternate option)."

First of all, you're of course forgetting miscarriages.

Secondly, you're forgetting that the births would be recorded in the state of birth, while the abortions would be recorded in the state of the abortion. That makes your whole calculation screwy.

Wes, what botched abortions are you referring to? ...

2009-06-03T15:39:41.598-04:00

Wes, what botched abortions are you referring to? There is no reference to botched abortions or any other abortions in your 13-year old reference.

Some current United States statistics:

*Childbirth maternal mortality rate is about 110 dead women per million childbirths. United States maternal mortality rate is ranked at 41st in the world. Bosnia! Croatia! and most of Europe, and Canada, Japan, and Australia have lower maternal mortality rates than our wealthy nation and its state-of-the-art health care. [1]

*Abortion maternal mortality rate is about 1 dead woman per million first trimester legal abortions. 90% of all abortions are first trimester. 1% are third trimester and are virtually all due to fetal abnormality or grievous maternal/fetal health risk. [2]

Childbirth is 110 times more dangerous to women than first trimester abortion. This is one (but not the only) really good reason why a woman should have the final say concerning her own body and her own health decisions. And this is also a good reason why John McCain should stop making air-quotes around women’s “health” concerns. And why you should go give your mother a big hug.

If we were serious about reducing the incidence of abortion, cleaning up the childbirth maternal mortality rate would certainly be a good place to start. Unless… maternal mortality might be an acceptable risk since women are only walking incubators anyway and are expendable once their job is finished ; ). Which begs this statistic: United States ranks 45th in infant mortality worldwide. Cuba! Slovenia! Czech Republic! rank better than us. In fact, the US infant mortality is ranked *second worst* in developed nations. [3, 4]

OT Fun fact: Viagra mortality rate is 49 dead men per million Viagra prescriptions. Hey John McCain, air-quote this! [5]

-----
1 World Health Organization, http://www.who.int/making_pregnancy_safer/topics/maternal_mortality/en/index.html
2 Guttmacher Institute, http://www.guttmacher.org/pubs/fb_induced_abortion.html
3 Central Intelligence Agency, https://www.cia.gov/library/publications/the-world factbook/rankorder/2091rank.html
4 CNN, http://www.cnn.com/2006/HEALTH/parenting/05/08/mothers.index/
5 Journal of American Medical Association, http://jama.ama-assn.org/cgi/content/extract/283/5/590?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=&fulltext=viagra+mortality&searchid=1&FIRSTINDEX=0&resourcetype=HWCIT

You should look for correlations in this data: ht...

2009-06-03T15:13:38.741-04:00

You should look for correlations in this data:

http://www.cdc.gov/mmwr/preview/mmwrhtml/mm4823a3.htm

Frequently, maternal mortality is associated with botched abortions.

Maybe pro-life states have lower rates of abortion...

2009-06-03T14:09:17.666-04:00

Maybe pro-life states have lower rates of abortions because they make it MUCH harder to get an abortion.

As several people pointed out, what matters is the rate of teen and other pregnancies per thousand.

Even this data is suspect because it is self reported and in conservative (anti-abortion) states there is a tendency to cover up and hide teen pregnancies.

"liberal paradigm": facts instead of ideology

2009-06-03T09:48:20.634-04:00

"liberal paradigm": facts instead of ideology

@Ole Forsberg No. 1) there is no way to estimate...

2009-06-03T09:10:19.212-04:00

@Ole Forsberg
No.
1) there is no way to estimate the exact number of pregnancies. Live births are reportable by law, and infant deaths are reportable by law. Fetal deaths are reportable around 20-24 weeks in most places. There is no reported data on lost pregnancies before the reportable gestational age, and abortions are not reportable by law in most places. So we have no denominator available to calculate a rate with pregnancy as the denominator. Virginia tried to pass a law to make all lost pregnancies reportable by law regardless of gestational age a few years ago, but it wasn't passed.

b) Nate has _tried_ to be clear about the limitations of the data, but he doesn't understand the data that he has. Not only does he not have a valid denominator, but he cannot link the available data to mother's state of residency, and thus the whole premise is invalid.

An example: Kansas and Colorado, apparently have the only two clinics that provide late-term abortions. Women from all over the US go to Kansas for the procedure. (Andrew Sullivan has been incredibly generous to facilitate the discussion.) By Nate's analysis, he would say that Kansas, although one of the most pro-life states in the US has the highest "rate" of late term abortions in the country. Do you see the problem there?

Your other observations, and I am assuming you are sincere, are equally flawed. Guttmacher doesn't have the data because there is no data on state of residence. That analysis _can not_ be done. There _is no data on that_. Nate may be good, but he isn't a magician.

Your third to last paragraph isn't right. In order to calculate population measure, you must know the exact denominator. This isn't a survey, okay? but even surveys have a population as a denominator, based on the census or some other known measure. Bu

But neither pregnancy nor abortion are reportable conditions. So it is impossible to calculate an abortion "rate."

Finally, please argue with my points, which are based on biostatistics and methodology, instead of snide irrelevancies about "liberal paradigms" whatever that means.

I am fascinated by the hypocrisy of some anti abor...

2009-06-03T07:41:51.964-04:00

I am fascinated by the hypocrisy of some anti abortion campaigners (I refuse to call those peolpe I am talking about here pro-life). How can people claim to be pro-life whilst glorying in murder? That disgusts and offends me. Whatever you think about Dr Stiller's work, surely his murder is as abhorrent.

That said I wonder if the left wing talk shows, especially Olberman and Maddow, were pushing the 'terrorism that works' meme just a touch too hard. Is there much evidence of doctors who would like to perform abortions but are too scared to?

Personally on that issue I think its just as likely that the kind of person who wants to become a doctor may find it very difficult to reconcile that instinct with the act of abortion, in most cases of the procedure.

@Tom: 1) ""The data does NOT represent the number...

2009-06-03T04:54:23.275-04:00

@Tom:

1) ""The data does NOT represent the number or rate of pregnancies that end in abortion.""

The number of pregnancies equals the number of abortions plus the number of live births less the number of failed pregnancies (unless I am forgetting an alternate option). The US (per the CIA) has an IMR of 6.26 deaths/1,000 live births, which is small compared to both the number of abortions and the numbers of live births.

b) Nate is _very_ clear vis-a-vis the limitations of the data. Do you have a better source for the data that Nate should use? I could not locate a data set at the Guttmacher Institute website (except for the one that has four data points per state for a 15-year period, and a discussion of their data on page 7, which is as sound as the CDC data). Furthermore, Guttmacher Institute does not appear to correct for your objections, as it only covers 46 states, provides only the distance traveled, and does _not_ state the state of origin for the woman (for which you excoriate Nate): ""Many state health departments are able to obtain only incomplete data from abortion providers, and in some states, only 40–50% of abortions are reported. Nonetheless, we sometimes found the information useful even in states with incomplete reporting."" (Source: Page 7, footnote) Incomplete reporting?

Furthermore, ""Of the 2,310 facilities surveyed, 916 responded"" (Source: Page 8, top). Including the entire number of responses, 1552/2310 gives a return rate of 67% (which is actually _extremely_ good for surveys).

Nate has stated quite clearly that those US states with a higher rate of pro-life residents have a lower rate of abortions in the state. This is clearly supported by the caveats you (and Nate and the CDC and the Guttmacher Institute) provided.

If reality does not fit the Liberal paradigm, I'm sorry; it bothers me, too.

You have some serious flaws in your analysis, Nate...

2009-06-03T01:56:23.714-04:00

You have some serious flaws in your analysis, Nate.

1) The data does NOT represent the number or rate of pregnancies that end in abortion. Instead, you are looking at a) the reported number of abortions per 1,000 live births (a ratio) and b) the reported number of abortions per 1,000 women aged 15--44 years. (an estimated population fertility measure.) This isn't even close to being what you describe as an "abortion rate." There is no way to measure the rate of pregnancies that end in abortion, because pregnancy is not a reportable condition thus there is no valid denominator for an "abortion rate." Live births is a reportable condition, so you are looking at the ratio of reported abortions to live births. And then they are reporting the population measure of reported abortions per population of women of childbearing age. Neither of those measures is anything like "number of pregnancies that end in abortion."

b) Under the Limitations section, it says "abortion data are compiled and reported to CDC by the central health department in the reporting area in which the abortion was performed rather than the area in which the woman resided." Therefore, any attempt to link that to state-based data is invalid on its face. You cannot correct for that with the available data. You need state of residence data for your analysis, and it isn't available.

I recommend that you read the entire Limitations section, Nate, and I think that you cannot support your opening thesis at all, i.e., "States with higher numbers of adults who identify themselves as "pro-life" have lower rates of abortion among both teenage and adult women." Indeed, the entire analysis is profoundly flawed and of very little use overall, even to look at trends over time, given that in some places the numbers provided are estimates in some years and reports in other years, some states don't report at all, the numbers have an acknowledged underreporting of anywhere from 12% to 20%.

I generally respect your work, Nate, but it isn't very helpful when you produce such a profoundly flawed analysis like this on such a hot-button topic. Please rethink. You don't want your flawed conclusions, I would hope, to be echoing around the blogosphere. Please consult a non-partisan epidemiologist to help you craft an analysis, preferably one who specializes in maternal child and perinatal health. Start with data from Guttmacher Institute. They have more consistent surveillance so their data may be helpful to look at trends, even with an underestimate in reporting. But I doubt even then you could do your analysis, because there is probably no valid state of residence data to be had.

@ToddVierling: "it pains me to know that there ar...

2009-06-03T01:39:44.751-04:00

@ToddVierling: "it pains me to know that there are so many children out there needing homes while the "pro-life" groups are now on an adoption-promotion kick."
Except of course that these same people are out to forbid same-sex couples from adopting (same-sex couples are a small percentage of all couples, but disproportionately represented among couples seeking to adopt).

When US anti-abortionists treat all life with th...

2009-06-03T01:18:51.856-04:00

When US anti-abortionists treat *all* life with the deference they purport to give american embryos I'll take them seriously enough to listen to their arguments. As it stands, they are just full of shit.

Nate, Good analysis, and good caveats (which are...

2009-06-03T01:15:23.345-04:00

Nate,

Good analysis, and good caveats (which are often more important in quantitative literature than anything). You are starting to write more like Prof. Gelman: provide the evidence, state the caveats, and suggest an appropriate conclusion.

I have noticed that several people commenting have asked for a different analysis than the one you did. Perhaps you could provide your data to them so that they can do their own regressions with their own models and their own assumptions, so that they can publish their individual findings on this site (as opposed to you doing all their work).

Also, could you link to your .do file (if you use Stata), or your .R file (if you use R), or cetera. This would allow us to learn how best to do the analyses for ourselves. I am not sure how this would be best accomplished within the confines of the Blogger software.

I was lucky. Many others aren't. In my book, bring...

2009-06-02T23:04:26.714-04:00

I was lucky. Many others aren't. In my book, bringing a child into a home that can't provide decent support, or throwing a child to the wind because of religious beliefs, is a far more cruel thing to do than terminating a pregnancy.

I still fully support abortion in spite of being adopted myself. You "pro-lifers" don't want abortion? Fine. Go adopt enough unwanted children to make the foster care population ZERO, and then we'll talk.

Hypocrites, all of them.

You seriously believe that any child would rather be dead than live in foster care?

I agree with you, judas. I thought the post was qu...

2009-06-02T20:09:19.659-04:00

I agree with you, judas. I thought the post was quite interesting, whatever its limitations.

Adrian: "Terrible analysis. What people actually c...

2009-06-02T19:56:11.852-04:00

Adrian:
"Terrible analysis. What people actually care about is the per capita absolute number of abortions.

translation: "This isn't the research I was interested in. "

Chris Milroy posted: "[...]I would sure think few...

2009-06-02T19:36:00.931-04:00

Chris Milroy posted:

"[...]I would sure think fewer abortions happen in marriage than out of it."

If you mean that a larger percentage of pregnancies out of wedlock would end in abortion, that does seem intuitive. If you mean that in absolute numbers, most abortions are to unwed mothers, I wouldn't assume so. I seem to recall at some point that pro-choice organizations actually marshalled data to the effect that most abortions were to married women, though I don't know if that used to be true and no longer is and would love to see current data. But there are, of course, many reasons for a married woman to have an abortion, including contraceptive failure (e.g., condom broke or came off inside), medical complications to her or/and the fetus, sudden divorce or/and loss of income, and rape.

Scott L said... Didn't some group do a survey on a...

2009-06-02T17:57:53.194-04:00

Scott L said...
Didn't some group do a survey on adolescent girls that attend christian schools and the result showed the opposite correlation than Nate's analysis? I seem to recall that it was released fairly recently.

Scott,

I also saw that - I was thinking an article on the study had been published this past weekend. I just spent a few minutes trying to find a link to the article and/or study. So far, no luck.

Mike in Maryland

My Blogger ID is http://www.blogger.com/profile/02848893412251095965

Comments on FiveThirtyEight: Politics Done Right: Pro-Life States Have Lower Abortion Rates

Looks like some very fishy "science going on here ...

酒店經紀人, 菲梵酒店經紀, 酒店經紀, 禮服酒店上班, 酒店小姐兼職, 便服酒店經紀, ...

@Tom: Great post on the weaknesses of the data. ...

Theonlysaneone: Making abortion illegal would eff...

theonlysaneone: This data, while interesting, does...

Not a very responsible post Mr. Silver. The data ...

@theonlysaneone: "You seriously believe that any c...

I am pro-life so please don't take this the wrong ...

Ole Forsberg -- you said "The number of pregnancie...

Wes, what botched abortions are you referring to? ...

You should look for correlations in this data: ht...

Maybe pro-life states have lower rates of abortion...

"liberal paradigm": facts instead of ideology

@Ole Forsberg No. 1) there is no way to estimate...

I am fascinated by the hypocrisy of some anti abor...

@Tom: 1) ""The data does NOT represent the number...

You have some serious flaws in your analysis, Nate...

@ToddVierling: "it pains me to know that there ar...

When US anti-abortionists treat *all* life with th...

Nate, Good analysis, and good caveats (which are...

I was lucky. Many others aren't. In my book, bring...

I agree with you, judas. I thought the post was qu...

Adrian: "Terrible analysis. What people actually c...

Chris Milroy posted: "[...]I would sure think few...

Scott L said... Didn't some group do a survey on a...

When US anti-abortionists treat all life with th...