Friday, July 16, 2010

Lies, Damn Lies, and Statistics

In an op-ed piece entitled "Open birth records will mean fewer adoptions," the National Council for Adoption argues that New Jersey's new legislation to open birth records will lower adoption rates just as, they claim, it did in other states that passed open records laws:
Oregon passed its law in 1998, after which there was a brief spike in adoptions, from 320 in 1997 to 403 in 2002. However, in 2007, the number of adoptions declined to 273, an 18 percent decrease in adoptions since Oregon's open-records law was passed.

Alabama enacted its law in 2000. In 1997, there were 250 adoptions. A brief spike occurred in 2002, when the number of adoptions rose to 370, but in 2007, that number declined precipitously to 124 adoptions, a 50 percent decrease from the number before the law was passed.

New Hampshire had 114 adoptions in 2002. In 2005, its open birth records law went into effect, and in 2007, the number of adoption declined to 90, a 21 percent decrease in adoptions after the law took effect.

Therefore, rather than promoting adoptions, the law statistically has had the opposite effect. Does New Jersey really want to risk a reduction in the number of adoptions? Is it any wonder that only six states have passed such laws?
When I first read the NCFA op-ed piece, I laughed out loud. The statistical picture was so obviously incomplete, and quite clearly illustrated the fallacy non causa pro causa, thinking that correlation equals causation. Sure, adoption rates decreased in states that passed open records laws -- because adoption has decreased in ALL states! It has decreased in states that have open records laws, in states that don't have open records laws, in states that have never had closed records at all! Newborn/infant adoption has been decreasing yearly in the U.S. That doesn't mean that open records has anything to do with it! We might as well say that the decline in adoption is caused by increased obesity rates, since they've occurred at the same time.

The Evan B. Donaldson Adoption Institute has recently released an update of their 2007 report of the effect of open records laws on adoption rates.  They explicitly call out the NCFA on their statistics:

The Adoption Institute requested the full study from which the three states’ statistics were drawn to check the accuracy of the conclusions and, most important, to ensure they were presented in context – especially since NCFA in the past has been accused of manipulating data in order to further itsaims. NCFA refused the Adoption Institute’s requests, however, and said it would provide no further information until it releases its full report next year. As a consequence, until then, the data in commentary will be the only information available to lawmakers or others dealing with the issue, but without context or opportunity for independent analysis.
The Donaldson report goes on to say, and illustrate in table form:

The rates of infant adoption per 1,000 abortions in Kansas and Alaska, where adult adopted personsalways have had access to their OBCs, are much higher than the national average. Both Kansas and Alaska have higher rates of adoption per 1,000 non-marital live births than the national rate. Adoption rates vary markedly from state to state. Of those in the table above that restored access prior to 2002, two states had adoption rates higher than the national average and two had lower ones. In comparing adoption rates in five states with access (Kansas, Alabama, Delaware, Oregon and Tennessee) to bordering states without access (Nebraska, Georgia, Pennsylvania, Washington and North Carolina), those states with access had higher adoption rates.
The statistical picture is far more complicated and nuanced than NCFA wants us to believe.  Read the whole Donaldson report;  it does a good job of explaining the importance of open records to all members of the adoption triad, the current state of the law, and the statistical picture.  Don't fall for NCFA's lies, damn lies, and statistics.

No comments: