Some statistics take a long time to get the true value. But we want a number now. So statisticians estimate.
Let me give two examples to explain what I mean.
Fertility rate is the number of babies that the average woman has in her lifetime.
If you wanted to know the true fertility rate of all females born today, you would have to wait until they had all lived their entire lives and died, and then count how many babies each woman had. But that would mean waiting for decades, and we want a number now. So what statisticians do instead if count the number of babies born to 1 year old girls (presumably zero) divided by the total number of 1 year old girls, plus the number of babies born to 2 year old girls divied by the total number of 2 year old girls, etc. Presumably we'd start to see some non-zero numbers around 12 and big numbers around 19 or 20, but that's not the point. We then add up all these numbers and we get an estimate for the lifetime total. To make a simplified example, suppose 50% of the 19 year olds have a baby, 40% of the 20 year olds, 20% of the 21 year olds, and that's it. No one under 19 or over 21 has a baby. Then the lifetime total for the average woman would be 0.5+0.4+0.2=1.1.
But there's a huge catch to calculating this way. It's very vulnerable to short term changes in rates. Suppose that women start having babies later in life. Maybe economic problems lead some women to delay having babies until they've worked longer and saved up some money. Maybe medical advances make it safer to have babies later in life. Whatever. If many women decide to delay having their next baby for a year or two, the birth rate would go way down the year that happened, and then go back up. Even if women have the same number of babies in their lifetime, if when they have those babies changes, the statistics can be thrown off.
There's also a techincal problem: Statisticians count the number of babies born to the number of women of a certain age alive today. But (to put it bluntly) some women die before they can have all the children they might have had. Instead of counting the number of babies born to the number of women who are, say, 20 years old, we should really be counting the number of babies born to women who were born 20 years ago. The difference being, we should count zero babies for each dead woman. But no one is keeping the statistics this way. If we counted correctly, a fertility rate of 2.0 would result in the same number of people in the next generation as there were in this generation: each woman must have one baby to replace herself and one to replace a man. (Assuming that exactly half of children born are boys, which is not quite true but close.) But given the way statistics are actually calculated, we really need a calculated fertility rate of about 2.1. Technically, higher in places where more women die young and lower in places where women die old. But 2.1 is a good working number.
There's a similar problem calculating the divorce rate. If you wanted to know the true percentage of marriages made today that end in divorce, you would have to wait until every one of those marriages ended, either in divorce or death of one person or the other. But that would mean waiting decades. We want a number now.
So what statisticians do is count the number of couples who marry and compare to the number of couples who divorce today. If, say, 100 couples get married and 40 get divorced, they say there are 40 divorces for 100 marriage, or 40% of marriages end in divorce.
But this is comparing two very different things: the number of single people who get married, and the number of married people who divorce. That is, you are comparing the number of marriages from a pool of single people, to the number of divorces from a pool of married people.
Think of some extreme cases. Suppose that in a certain area there are 10,000 married couples. This year, 100 of them get divorced. In the same year, there are 100 marriages. So there are 100 divorces and 100 marriages. Would it be fair to say that the divorce rate is therefore 100%? That every marriage ends in divorce? Surely not. Or go even further. Suppose that there are 100 marriages but 110 divorces. That is certainly mathematically possible. Does that mean that 110% of marriages end in divorce? That if you get married, there is a 110% chance that you will divorce? Of course that's nonsense.
Suppose that some social or political or economic change makes marriage less popular. More married people get divorced and fewer single people get married. The calculated divorce rate could soar. For example, suppose there are 10,000 married couples. This year, 100 of them get divorced while 222 new couples get married. The calculated divorce rate is 45%. Next year 90 couples divorce while only 100 get married. The calculated divorce rate is 90%! Marriage is dead! But while the divorce rate doubled, the percentage of couples getting divorced actually fell, from 100 out of 10,000 = 1.0% to 90 out of 10,132 = .89% .
Of course predictions based on statistics like this depend on assuming that present trends continue. This is often a wildly unlikely assumption. In 1949, Americans were buying 100,000 televisions a week, or over 5 million a year. At that time there were 150 million Americans, so if that trend had continued, by 1980 there would be more televisions in the US than people. by 2010 there would have been 2 TVs for every person. But of course that trend did not continue. in 1949 almost no one owned a television, so people were "catching up". Once someone owned a television, they were far less likely to buy another. Someone who has no TV is far more likely to buy a TV than someone who already has one. He might buy a new one to get a better model or to replace a broken one, or to have a second TV for the bedroom, or whatever. But from zero to one is more likely than from one to two.
Likewise, in an era when few people are having babies, the social value of having a baby is likely to go up. The couple with a baby is helping to preserve civilization. There is less competition for things needed by children, like baby formula and schooling, so costs likely go down. And governments are likely to implement policies to encourage having babies. With more incentives to have babies, the number will likely go up. Will it go up enough to make a difference? We'll see.
© 2026 by Jay Johansen
No comments yet.