A friend of mine used to tell the joke about a man standing on the corner, clapping his hands continuously. A passer-by asked him why he was clapping his hands – “To keep the elephants away.” The passer-by said, “But there aren’t any elephants here.” The man: “See, it’s working.”

What causes a change in the value of a statistic is often left to similar speculation.

We had an illustration of that in a recent news story about the 2009 American Community Survey estimates showing a decline in the number of persons currently married. The story asserted that the decline in the number of marriages was directly related to the current economic downturn.

Many factors can affect the estimates of the number and proportion of people currently married. For example, declining numbers could reflect the passing of members of an older generation that had higher marriage rates. A look at long term trends confirms that the proportion of people who are married is on the decline in this country, but it has been for decades (this can be calculated from 1950 using the Census Bureau’s Current Population Survey, historical time series table MS-1).

The impact of an external event, such as an economic recession, can’t easily be teased out of the change over time. It would be useful to such inference to see whether persons considering marriage before and after the recession were making different decisions. It would be useful to know whether those couples most affected by the recession (e.g., losing a job, having a home foreclosed), were more prone to put off marriage relative to those unaffected by the recession. But these estimates were not part of the ACS report.

Statistical estimates are critical to understanding our nation, who we are and how we live. We just need to take care that we understand what they can and cannot tell us about our country.