Rachel Lehmann-Haupt wrote in Newsweek 11-18 May 2009: "Egg freezing, I believe, could be as revolutionary as the birth-control pill. And the timing for its takeoff couldn't be better. The age of first-time motherhood is rising. In the United States, the number of women becoming pregnant between the ages of 35 and 44 has nearly doubled since 1980. As education, advanced degrees and higher salaries become priorities, we are trading in our years of procreative power to gain economic power."
I tracked down the statistic the author references (thank you eppc.org for actually referencing the stat) in a press release from the National Center for Health Statistics. The press release states, "The birth rate for women aged 40-44 years has more than doubled since 1981." It also states that the birth rate for that age group in 2003 was 8.7 births per 1000 women. I had to dig through their archives to find the exact birth rate for 1981 (it was 4.0 for all women 40 and over).
The data does indicate a steady increase in the birth rate for women over 40. However, when the birth rate for all the other age categories are included in the plot, the tend is. . .underwhelming.
As you can see, the statistic is entirely accurate, but it is also completely out of context. Until the birth rate of the 40+ age category doubles from 8 to 16, and then again to 32, it will still be a relatively insignificant event. Mentioning that it doubled is like mentioning that America is using more wind power. Yeah, we are, but even "twice as much" still doesn't matter. In fact, the most significant increase seems to be in the 30-40 age range, not in the 40+ age range.
After looking at the data (in what is admittedly a quick and dirty fashion) it appears the author's main point seems to stand. Overall the birth rate seems to be decreasing and, at the same time, the birth rates for older women have been increasing. From a systems perspective, this is an interesting trend. What factors could be driving this change? Is it "education, advanced degrees and higher salaries" as the author states?
One of the things to keep in mind when looking at a system is the difference between correlation and causation. Just because two metrics change the same way at the same time does not mean one is affecting the other. For example, an increase in average global temperature coincided with a decrease in average global pirate attacks. That does not mean that pirates are allergic to heat, and it does not even mean they are both responding to a change in some third factor, it just means that they both changed.
I didn't write this to get into the debate. I only intended it as a small case study to illustrate a point. Statistics should be very carefully applied to systems. Statistical analysis is a great tool for reductionism, but it is less useful for analysis of holistic systems. This is due to the fact that statistics, by necessity, can only be used when things are reduced to specific metrics. It is tempting to analyze every subsystem and think that you understand the system, but to understand the system you have to analyze the system itself, not the subsystems. Doing this properly requires a lot of careful definitions of exactly what the system you are studying consists of, which are pretty boring, so that step usually gets skipped (or at least not mentioned).
When you want to talk about a system, but you skip the definition step, what you say probably doesn't matter. At the very least, it is open to a lot of misinterpretation. For example, in the case I cite here the author could have been more specific about what the stat(s) were actually describing, how they were obtained, etc.