Question - does anybody know if there's a way to get TimesSelect coverage if you aren't an individual subscriber to the paper? We've (at work, in a library) found that Stanley Fish's columns, for example, aren't in the print edition, and don't make it into Lexis or Factiva. The library has a print subscription, but it's institutional and doesn't include TimesSelect (it's not an option).

I'm not surprised about the median figure because there's a much larger group of marriageable people on the older side of 20 than the younger side.

That's why I'm wondering if there's another reason they're using median instead of mean.

Is there a particular advantage to looking at the median, rather than the mean, in a table like the one linked above?

The distribution isn't bell-curve shaped. There will be quite a few people getting married in their twenties, and it will tail off as people get older. While there may be only a few people getting married in their 90s, it will make the average much higher than the median, and misrepresent the data.

My guess (ex-cloaca) is that in 1890 people waited to get married until they could afford to set up their own households, i.e. in rural areas. It took a while to amass the cash to do that.

What I found interesting (and comforting I guess) in the marriage stats was the huge jump in the "Never Marrieds" found here: Percent Never Married 1970-2004

Is there a particular advantage to looking at the median, rather than the mean, in a table like the one linked above?

I think it guards against outliers, so someone marrying for the first time at 50 doesn't throw it off as much as it would in a mean. If you have 15, 20, 20, 20, 25, 25, 50, the median is 20, but the mean is 25.

Okay, that makes sense. Well then, I want all these tables in median, mean, and mode.

I also want a cookie for remembering 'mode'.

I want a pony.