15-year-old “women without husbands”: reporting on statistical analyses

In February of 2007, The New York Times (and, I am sure, many others) reported on the most recent findings from the census bureau, which noted we had passed a milestone, of sorts, in that over 50% of American women were now living without husbands.

Critics of the Times quickly discovered that the count of “unmarried women” reported by the demographers included girls as young as 15 living with their parents and pounced on that as evidence of a cultural (progressive) bias in the reporting.  FOX News ran with it, in its typically hyperventilated way, and the predictable outrage followed.  In response, the Public Editor for the Times (Byron Calame) penned a review of the reporting and editing process in which he cited what he described as “several journalistic lapses” which both inflated (slightly) the statistic and (because of the magic in phrase the “over half” that might not have been associated with some different phrase, like “nearly half”) elevated the story to front-page status.

I wrote the following letter to Mr. Calame, not because I found his specific critique wanting (I have generally found the Times‘ Public Editors to be both fair and willing to be critical of their own reporters and editors when that is called for) but because I thought the subject was broader than the specific issue of whether or not 15-year-olds might legitimately be described as “women” for demographic purposes.  Alas, something else (which probably seemed very important at the time and now leaves no trace in my memory) distracted me before I actually sent it, so it never saw the light of day.

11 February 2007

I read your take on the controversy about “15 year old women” living without husbands with both delight – at the fact that such journalistic sloppiness was called to account – and dismay – at the narrowness both of your critique (whether or not specific age groups were or were not counted in the statistics) and of the response by the editors (that they would work to create a “vetting network to help with the editing of articles dealing with those subjects [statistics and demographics]”).

I am an engineer.  My work often involves statistical analysis of large data sets from biological and, occasionally, from demographic sources.  I am intensely interested in public policy.  From that perspective, I can say that the general quality of reporting on issues involving statistical analysis – from social sciences to ecological issues (including global warming) to medical and epidemiological studies to economic projections – is appalling, not just in The New York Times but across the whole range of media.

Much of such reporting is not merely clumsy but laughable – it doesn’t even pass a first-order smell-test.  The Boston Globe once reported uncritically on a particular researcher’s claim to have established a mathematical relationship between IQ and SAT scores – one which, on inspection, predicted that SAT scores should hit a perfect 1600 at an IQ of 121 and decline thereafter as IQ increased.  Perhaps smarter people get bored.  The San Jose Mercury News reported a few years ago on a study from the NIH which, they claimed, showed a dramatic decline in lung cancer as a result of California’s anti-smoking initiative – a decline which began the very year the initiative was launched and proceeded linearly downward thereafter.  Even the most fervent believer in the magic of government programs can’t honestly assert that people will stop getting lung cancer the instant the public service ads hit the air.  And, as I mentioned in a letter to your predecessor in another context, a New York Times headline once proclaimed that “Cancer Passes Heart Disease as Top Killer”, only to reveal in small print buried near the bottom of the accompanying story that the death rate from both had actually declined, but that deaths from heart disease had declined more dramatically.  If only all bad news were so good.

But, although I agree wholeheartedly that there is a general ignorance among both reporters and editors (and, by the way, among lawyers, and politicians, and “activists”, and a lot of humanities professors, and even some social scientists, and ecologists, and economists, and physicians) about those subjects, and that much better vetting of such stories is essential, the real problem goes beyond those narrow bounds.

The broader problem is that statistical and mathematical and scientific ignorance is magnified by a fundamental disconnect in language between those who generate such information and those to whom it is reported.  Sometimes that disconnect is the result of simple misunderstanding.  But often it is intentional – a kind of rhetorical bait-and-switch used to manipulate public opinion.

Consider, first, the subject of your article: women living without husbands.

Demographers have specific and technical classifications they use – by age, by living arrangement, by financial category – from which they extract meaning and understanding about populations and population statistics.  They may use a shorthand term like “women” to group categories for their own purposes, but that doesn’t mean that their use of the term corresponds, in any strict or meaningful sense, to the broad understanding of that term by anyone outside their academic specialty.

To the general public, the phrase “women living without husbands” communicates something beyond mere age or even simple living situation: it implies

Women who we might otherwise expect to be married but who, for some reason, are not.

Note that this does not necessarily reflect a literal and precise reading of the phrase.  Rather, it reflects the interaction of the language with the cultural assumptions we bring to our understanding of it.  Clearly, in that context, the teenagers living with their parents who were the subject of your ire would be excluded from that category, but most people would also subliminally exclude teenagers living on their own (they are too young to marry responsibly anyway – unless they are having babies which changes that equation), anyone still in school (we don’t necessarily expect them to marry before they finish), people who are engaged to be married or otherwise actively headed toward marriage, anyone who had been recently widowed (within the last several years) or widowed in old age after a long marriage, married women whose husbands are temporarily away from home (you mentioned these), married women who are themselves temporarily away from home, and, perhaps, even women whose long-term and otherwise stable relationships would traditionally have constituted a de facto “common law” marriage even if they were never legally formalized.  In other words, what we understand from that phrase is really:

Women living outside of our expectations about marriage.

If a reporter had used the phrase “women living on their own”, or “women who choose to be single”, or some other less (or differently) culturally-loaded formulation, they might have communicated something closer to the meaning the demographers intended.  But, as phrased, they have – intentionally or merely carelessly – set up an expectation in their readers not supported by the facts.  At that point, their responsibility goes beyond explicating a few details of the data set; they must strive specifically to put all proper bounds on that expectation.

A similar disconnect occurs when we talk about “children” who are victims of violence or, in the more popular narrative, “children who have been killed by guns”.  The phrase brings to our minds visions of kindergarteners mowed down in the street or shot by their playmates with daddy’s carelessly-stored Saturday Night Special.  Coupled with the latest statistics, we get a picture of wide-scale havoc, of innocence betrayed.

But, although that does sometimes happen, the real story of these “children” is that the great majority of them are in their late teens, on the cusp of adulthood, are active in gangs, and are as likely as not to have gone down with their own guns blazing.  That story represents a problem in its own right.  But it is not the picture of innocence betrayed that the phrase conjures in our minds.

© Copyright 2007, Augustus P. Lowell

Leave a Reply