Past predictions (marriage dreams not quite true)

Thoughts for the demographer in all of us. And a shout-out to those demography readers, if any, who enjoy predictions, projections, and forecasts.

Demographers have a very useful way of imagining the future as a quality of the present. Because the present always implies a future (and here I think you could add “going forward” without being completely redundant). This imaginative act may be mathematically accomplished, but — if you can feel it — it’s as emotionally compelling as gripping dystopian fiction.

The most common uses of this technique are life expectancies and total fertility rates. For these calculations, demographers line up current events by the age of the population in a spreadsheet (deaths by age or births by age), and press PLAY to produce a future based on constant repetition of the present. So they’re not really telling me I’ve got 35 years left — they’re telling me that if I lived 2006 over and over again as I age — 100,000 times — my average self would live 35 more years.

Looking at things this way gives a forward-looking description of the present, but it also sometimes actually does a good job of predicting the future. However, the prediction only works when things aren’t changing radically. So if they cure cancer and heart disease in the next few years, all of us 40+-year-olds will get a new lease on life expectancy.

And, on the fertility side, if everyone decides to put off childbearing for a year or two because of the recession, but then gets back on track once the Tea Party fixes everything, our fertility estimates will be way off for a couple of years.

Forecast: Higher education with a chance of marriage

Marriage is one of those things that’s hard to forecast because the dynamics of the institution are changing rapidly (in generational terms). So, we can tell you that — for the first time in spreadsheet-recorded history — never-married young adults (age 25-34) now outnumber those who are married. But we can’t tell you how many of them will eventually get married; that future has yet to occur, and, unlike mortality, is not certain.

In 2001, Joshua Goldstein and Catherine Kenney took a courageous stab at forecasting the eventual lifetime marriage rates for American Baby Boomers — people born between 1946 and 1964. Their paper in the American Sociological Review has been cited an even 100 times as of today, making it highly influential and worth reading. The two take-home points remain useful and true:

• About 90% of Boomers would eventually marry, though starting at later ages than in the past. (In fact, women born in 1946 have already surpassed 93% ever-married, by my calculations.)
• Paradoxically, lifetime marriage rates will be higher for college-educated women — although it might seem they need marriage less. This has also come to pass.

But their predictions are coming in quite low, especially for Black women.* I have no idea why, but their forecasts have Black women’s marriage rates falling even faster than they actually have. Here are the predictions from the paper — for lifetime ever-married rates — and the actual percentage married as of 2007 (based on 2006-2008 pooled samples from the American Community Survey, courtesy of IPUMS):

So, to interpret the graph, for example, their prediction was that 85% of Black women born in the late 1940s would eventually marry at some point. As of 2007, when those women were about 60 years old, that is exactly how many of them have been married. That wasn’t such a hard prediction, though, since most of those marriages took place before the data for their model were collected (1995).

Why are Black women from the later cohorts getting married at higher rates than predicted? I have no idea. But seeing as the youngest cohort they studied are only in their late-40s now, their prediction can only get further off (since people can never get un-ever-married).

Finally, to show the difficulty in predicting this kind of thing, consider the ever-married rates, by age, among my 2006-2008 sample — a simple cross-section of the current population:

Source: My chart from ACS data from IPUMS.

This shows you how different the Black population is from the other major race-ethnic groups, and it also shows how fast things are changing.** More than 90% of 80-year-old Black women have been married (although note differential survival may be an an issue), but only 50% of those age 33 have been married.

Technically, those 33-year-olds have 47 years to catch up to their grandparents — who got married in the 1950s. And that is one kind of prediction you could make: “Almost all of today’s 80-year-olds married, so today’s youth will, too.” But it seems unlikely. And since there is method — not just math — behind the demographic madness, we don’t do predictions that way. If you can figure it out, though, I’ll devote a blog post to you.

* Note that Goldstein and Kenney anticipated that, if their predictions were off, they were likely to underestimate rather than overestimate marriage rates.

** I say “population” even though the data are for women. That’s an old, sexist demographic convention. Marriage rates for men are pretty similar.