Lies, Damned Lies, and Statistics

[lollardfish]

The story of ever-increasing divorce is a powerful narrative. It is also wrong.

From the NYT.

Interesting piece.

Comments

[jadiana.livejournal.com]

Though I find it ironic that, an article about misleading conclusions of statistics, also sort of twists the numbers.

"The Census Bureau reported that slightly more than half of all marriages occurring between 1975 and 1979 had not made it to their 25th anniversary."

After accounting for additional data, the article states:

"Once these marriages are added to the mix, it turns out that a majority of couples who tied the knot from 1975 to 1979 — about 53 percent — reached their silver anniversary."

Ooo. Yeah, that's an astounding difference! Not. Note the use of "majority" when it's only a few percent. And the difference between the two calculations aren't, well, very different.

If 'slightly more than half means, say, 51% HAD NOT, then, what, 49% had? And now, recalculated, it's 53%?

Not that I don't agree that divorce is falling. I just have a hard time understanding why they made a big deal out of few percent, as if it were significantly different.

[kalmn.livejournal.com]

i am not a statistics geek, just related to one. however, a few percent is a huge deal, statistically speaking. that's enough of a difference (at this size of study) so that it is a significant difference, and not attributable to margin of error. if it were a hundred people, changing from "slightly less than half" to 53 percent would be attributable to somebody writing something down wrong. but since it's millions of people being studied, slightly less than half to 53 percent is a change of tens of thousands of people, if not more.

[jadiana.livejournal.com]

Yeah, I get that. :)


But my point was rather, that the article is saying that it's not as bad as all that, when really, it boils down to numbers that if you were to generalize, would still be called "about half".

The emphasis of the article implies a difference, that, frankly, isn't represented by a few percent.