Let's See If We Can Terminate This Meme

The Times writes about the Democratic Party's urge to find religion, and includes this hedged and under-sourced tidbit about abortion:

Citing statistics showing that the incidence of abortion fell under President Bill Clinton and rose under President Bush, they argue that the party can reach religious voters without flinching from its current stance on abortion rights by shifting the debate from the legality to the frequency of the procedure - a reprise of Mr. Clinton's formulation that abortion should be "safe, legal and rare."

Emphasis added.  The authoritative sources for national abortion statistics are the Center for Disease Control and the Alan Gutmacher Institute, neither of which has published a comprehensive study with data going beyond 2000.

The source for the belief that abortion rose under Bush seems to be a recent study by Glen Stassen, Lewis B. Smedes Professor of Christian Ethics at Fuller Theological Seminary, summarized here:

Using data from the Minnesota Citizens Concerned for Life, the Guttmacher Institute, and reporting by individual states, Stassen found that U.S. abortion rates declined 17.4% in the 1990s to a 24-year low when Bush took office. Many expected that downward trend to continue under the conservative president, but Stassen found the opposite: 52,000 more abortions occurred in 2002 than would have been expected under the pre-2000 conditions, and abortion has risen significantly in those states reporting multi-year abortion statistics.

The study was published in Sojourners, a liberal religious group cited in this story.

And my emphasis was added - the "52,000" figure refers to a deeply disputable baseline projection, which is one of many objections raised in the rebuttal offered by Dr. Randy O'Bannon and Laura Hussey.

Stassen responded to his critics, and they launched a second response.  My quick summary is the following:

Stassen's trend is misleading and/or non-existent - the abortion rate plunged after welfare reform, and was flat through the second half of the 90's.  Stassen's baseline projection that we should have expected 52,000 fewer abortions under Bush is very dubious.

He uses only 16 states with no evidence that they are typical of a national trend.

Stassen claims 11 of 16 states showed an increase; for two of these states, his data is simply wrong (he seems to have looked at births, rather than abortions, in one case); for two more, changes in reporting led the state health officials to urge caution in using the stats (Arizona, +26%; Colorado, +67% - it is hard to believe that is simply changes in access to health care or the economy that explains that).

Stassen also ignores inconvenient data - he has Illinois going up from 2001 to 2002; the data showing Illinois dropping 10% in 2003 was available when he did his work.

So, is the Times wrong?  Of course not - their statement that "Democrats are citing statistics" is factually accurate.  However, one might hope that they would note that the study itself is a bit controversial.  And one might hope that the "reality-based community" would be a bit more demanding in their analysis.

Eventually, the CDC and the Alan Gutmacher Institute will report, and we will have a more authoritative view of "the truth".  And regardless of what they report, a year from now Barbra Streisand will still be citing the conclusions of this study.  However, we would hope that the Times would provide some substantion of this claim, or warn us that the facts are in dispute.  The inclusion of the word "controversial" or "disputed" in the sentence describing the study would be enough.  Or , on a good day, they might even attach a name and a researcher to the study so that we can see for ourselves.

Daniel Okrent, Public Editor of the NY Times, could not care less about your concerns, but he can be reached at: Public@nytimes.com

For folks who want the full in-flight movie, and not the trailer, more of the back and forth is summarized in the continuation, with lots of links, starting with this story:

<p><blockquote>WASHINGTON (ABP) -- A scholar for the National Right to Life committee says an evangelical seminary professor's study suggesting that the abortion rate has risen under President Bush is flawed. But the professor is standing by his conclusion.

The analysis, by statistician and ethicist Glen Stassen of Fuller Theological Seminary near Los Angeles, began receiving attention after many news outlets ran an opinion column he wrote about it Oct. 12.

...One of their critiques was that Stassen's sample size wasn't large enough. "Since no national abortion data have been reported since 2000, Stassen looks at abortion figures for 16 states over 2001, 2002, and in some cases, 2003," O'Bannon and Hussey wrote. "Stassen confidently claims that abortions increased in 11 of those 16 states during the Bush administration and asserts that this reflects a larger national upward trend in abortions. Yet Stassen never demonstrates that his 16 states are representative of the 50 states."

Asked by an Associated Baptist Presss reporter for his response to the criticism, Stassen defended the sample size by pointing to statistics cited every day in the run-up to the presidential election. "The political polls include 500 or 1,000 respondents and extrapolate to 50 million voters. I have 16 states out of 50, which is a much greater sample, proportionately," he said.

I have to break in, here and applaud this "statistics made easy" approach.  A more convincing method might have been to take data from these 16 states during the 1990's, and compare that to the national trends. 

Those so inclined can ponder the notion that, since 16 is a higher proportion of 50 than 1,000 is of 50 million, the standard error might be less.  And since we glean this nugget from a fellow described as a "statistician and ethicist", we can also contemplate the ethics of misleading a hapless reporter about statistics.

O'Bannon and Hussey also noted that Stassen had incorrectly cited increases in the 2002 abortion rate in two states -- South Dakota and Wisconsin -- that actually showed decreases in their abortion rates. They also said that newer statistics on the abortion rate in Illinois showed a significant decrease in 2003.

Therefore, O'Bannon and Hussey said, "When one shifts Wisconsin and South Dakota to the decrease column, and adds in Illinois after its dramatic 2003 drop in abortions, Stassen's claim that abortions have increased in 11 out of 16 states now turns into an eight-to-eight tie, with as many states decreasing as increasing. Hardly anything definitive."

Stassen acknowledged the mistake with the Wisconsin and South Dakota figures, but said he did not use the Illinois figures because they were too recent and may be incomplete. He also said the pro-Bush researchers were engaging in the same crime of which they accuse him.

There were also serious questions about changes in reporting in two states:

Stassen reports large increases in four of the 16 states above – Colorado, Arizona, Idaho, and Michigan. While state data do record a significantly higher number of abortions in these states for 2002 than in 2001, officials from at least two of the states with the highest reported increases caution against seeing this as evidence of any real increase.

In Arizona, where Stassen reported a 26.4% increase, the state Department of Health Services cautioned in its report that “It is unclear whether this increase in the number of reported abortions represents a true increase in the actual number of abortions performed, or, perhaps, a better response rate of providers of non-surgical (so called medical) terminations of pregnancy.”

State officials in Colorado, where Stassen reported an astronomical 67.4% one year increase, recently revamped their reporting regimen to address underreporting, and sent a note to abortion “providers” reminding them that reporting was required in Colorado. The state said they expected an increase in reports, and declared, “No one could or should conclude that this anticipated increase in the rate of reported terminations reflects an increase in the true rate.”

Stassen doesn’t report these caveats. But if state officials are reluctant to say their data indicates real increases, they don’t belong on Stassen's list of states with more abortions. That would leave just 6 increasing versus 8 decreasing states, the opposite of what Stassen claims.

The misleading baseline calculation which led to a projected decline in abortions by 52,000 is discussed here:

Stassen looks at national figures showing the annual number of abortions dropping from about 1.6 million in 1990 to just over 1.3 million in 2000 to argue that there was a steady decline of 1.7% a year before Bush took office. There was a 17.4% decline over the decade, but Stassen’s claim is misleading. The decline was strongest in the first half of the decade, which began with George H.W. Bush in office, but slowed during Bill Clinton’s term, and even reversed itself one year. In Clinton’s last year in office, the decline was not 1.7%, but just 0.1%.

This report from the Alan Gutmacher Institute illustrates the point - there was a steep drop in the abortion rate up to about 1996, and then it leveled off.  Here is some CDC data.

MORE:  A Heritage Foundation study says that pro-life changes in state laws could explain the drop during the 90's.  I want to mention welfare reform.

Comments

[SomeCallMeTim]

Can we clarify at least one thing? Our big problem is our leadership. At least on enormous advantage that Republicans seem to have is that your party at least appears to be bottom-led. Your grassroots efforts are well-rooted; ours (at the moment) are just growing in.

All of which is to say that many Demcorats would consider it a favor if, when you read stories like the above, where you see "the Democratic Party" you read "the frigging morons at the top of the Party who, but for the personal skills of Bill Clinton, wouldn't have won a presidential election in a generation."

Thanks

[TM]

Good point.

[Andrew Grossman]

This paper seems relevant. It discusses a different explanation for the Clinton drop:

http://www.heritage.org/Research/Family/wm598.cfm

[TM]

Thanks very much. Welfare reform merits a nod as well.

[ed]

Hmmm.

I'd send an email or letter to Okrent but it's gotten old. I figured it was time to give up when all I could write in my last letter was "blah blah blah outrage yada yada yada bias". After you reach a certain point, long since passed, having an ombudsman is a complete waste of time.

I think the only message that would ever actually get read by Okrent would be one wrapped around a brick and sent through a window.

[max]

"Daniel Okrent, Public Editor of the NY Times, could not care less about your concerns"

He has gone native 100% hasn't he?

I read somewhere that his contract is not being renewed. If that's correct, I wonder what he'll be doing. Perhaps he's going to work for the nyt in some other capacity?

Also, does anyone know if his position as in-house cheerleader is going to be eliminated when his contract expires?

[Tim Worstall]

Just a thought, but are they counting actual numbers of abortions? If they are, are they correcting for the rise in the pop[ulation and ot changes in the age composition?

[Cecil Turner]

I looked this over briefly during the first iteration of the "Partial Birth Abortion" ban (trying to discern whose numbers were closer to reality), and concluded it was a hopeless muddle. No two sources agree on methodology or reliability, and there appears to be some merit to the various complaints:

The Guttmacher Institute (a special affiliate to Planned Parenthood) reports higher numbers than the Center for Disease Control (CDC) because Guttmacher actively collects the abortion data directly from providers. The CDC relies on figures gained from voluntary reporting by abortion providers and relayed from state health agencies. All numbers reported are voluntary; there are no laws requiring abortionists to report to any national agency the numbers of abortions they perform.

Starting in 1998, the CDC stopped including statistics from four states including California, a state that accounted for 23 percent of the nation's abortions in 1997.

AGI has estimated a possible 3-6% rate of underreporting. It's obvious from other AGI research that people routinely lie about abortions, and recent court cases have forbidden federal researchers access to individual records. Planned Parenthood clearly sees this as primarily a privacy issue, a view likely shared by AGI.

I'm extremely skeptical of any "trend" that shows up in the third significant figure of this data--even before factoring in changes in population. There might be something there, but I don't see how you'd know it.