We're all familiar with "scientific" studies and "objective" journalists distorting reality. Here are a few examples that happen to relate to the same-sex marriage issue we've discussed on these pages.
Under the heading 'A Breed Apart' the Opinion Journal's Best of the Web observes a mathematical oddity in a Boston Globe article about children of same-sex parents, and I added a few more calculations of my own.
The Globe states, "Virtually all the nearly 50 studies on the children of gay and lesbian parents--who number between 6 [million] and 14 million in the United States, according to various studies--have found no significant differences between children raised by heterosexual or homosexual parents." But according to a sidebar on the same article (near the bottom) the 2000 Census estimated there were about 659,000 gay couples in America, or less than 1% of households. These two data points imply that the average same-sex couple has from 9 to 21 children - according to various studies. If the Census figures are considered erroneous, and if we generously credit every same-sex couple with two children, then there are between 3 million and 7 million same-sex couples in America - according to various studies. If true, this would represent approximately 4 to 10 percent of all "family households." I suppose that's possible, but it would mean about 8 to 20 percent of "blue state" households (what with the "intolerance" of those in the "red states.") A more defensible estimate of the number of children of same-sex couples is probably less than one per 659,000 of the gay couples - a far cry from 14 million.
So much for the journalistic distortion, now for the scientific one. Later in the same Globe article is a discussion about whether ALL the results of a study should be reported or if some should be omitted or "lowballed ... for fear of how the findings would be used." The lowballed finding was that although being a child of same-sex parents apparently did not encourage children to "become homosexuals," those children were, in fact, more open to the idea and 24% of them (6 of 25) "had had a [sexual] relationship with someone of the same gender." The original researcher wrote in response to this "outing" of her interpretation that her critics had "failed to distinguish between children's attitudes, which can change, and gender identity, which is relatively fixed."
Several questions come to mind:
- Is gender identity fixed, or only "relatively fixed?" (Or moderately fixed, or appreciably fixed, or fixed except when it is variable?)
- Are children born with their gender identity or do they "become homosexuals?"
- Is it ok to use non-scientific methods for social science research since social science isn't really science?
Even more interesting is how many conclusions can be drawn by studying 56 out of shall we say 660,000+ children. So even taking the lowest estimate the study composed .01% of the studied population. I wonder how exactly the census estimates the number of gay couples? I don't remember my census form, is there a box for that? I have always thought that the gay rights folk who overestimate the number of gay people do themselves a great disservice to their cause. The number of 10% has never been believable.
Posted by: Silence Dogood at March 11, 2004 08:57 AMSome folks believe it. These statistics get repeated so often. As you point out, there are no credible baseline statistics. Lastly, there is such a general innumeracy among media consumers -- how many will multiply these figures out and see the error? We all remember high school algebra students looking earnestly at their calculators and saying "Yes, it would take 23,674.6433 hours for the ball to fall to earth from the 4th floor..." It is not hardwired into most people to question statistics.
My favorite study proved that 87.43% of statistics were made up on the spot. I have a link to the survey around here somewhere...
It does seem in this age of calculators and computers that the "sanity check" on numbers is forgotton. (Man, do I sound like an old fogey or what?) My college statistics professor began his first lecture by "proving" to us that wearing a hat causes cancer. You see, skin cancer is by far the most common form of cancer, and statistically bald people are more likely to get skin cancer, and statistically bald people are most likely to wear hats, so you can statistically prove that hat wearers have a much higher incindence of cancer.
Posted by: Silence Dogood at March 12, 2004 09:06 AMSilence is correct about the sanity check. My latest pet peeve is that kids are allowed calculators to take SAT's these days. I must be a fogey too.
However, the issue relating to hats and cancer is not a case of distortion of statistics but a problem with correlation and causation. By mixing them up, it is also possible to, "prove," that ice cream, "causes," violence when everyone knows that it is really hockey that causes violence.
Bad science is bad science in any field (social sciences included) and there is more and more of it these days. Kudos to John Galt to notice it and point it out.
Posted by: dagny at March 12, 2004 10:08 AM