Wednesday, July 20, 2011

Statistical Analysis Gone Bad

Before I launch into today’s rant about a very poorly written article, let me first acknowledge that I understand the reasons why this article was written and posted the way it was.

Internet “news” stories (and I use that term with great reservation) are ranked by the number of readers clicking onto a given story (aka “hits”). Hits are driven by catchy, sensational headlines. More hits = increased revenue.

I get that.

But it’s no excuse for blatantly and deliberately twisting statistics – in other words, lying – to get a reader’s attention.

I submit the following for your consideration. Please note: I am including the link to this article at the bottom of the post simply for the purpose of referencing the source. Read it if you wish, knowing that by doing so, you’re only increasing revenue for a story that should never have seen the light of day.

The headline of an AOL article published on July 6th:

“Let The Arguments Fly: Study Shows Women More Likely To Cause Traffic Accidents”

(Gee, nothing inflammatory or sensational there, eh?)

The subtitle:

“Researchers find women are more likely to get into accidents with other women, even though females drive fewer miles than men”

(Still with me?)

In the opening paragraph:

“A University of Michigan study… (finds) an inordinate number of accidents happen when both drivers are women.”

(Read that carefully.)

The research results as presented in the article itself:

Female-to-female accidents: 20.5% of all crashes.
Male-to-male accidents: 31.9%
Male-to-female accidents: 47.6%.

Go back and re-read that subtitle text that says “women are more likely to get into accidents with other women” and then look at the higher percentage – nearly twelve percentage points higher! – of male-to-male crashes.


What’s muddled in translation is the fact that researchers had expectations about the percentages of women and men involved in traffic accidents, and that the actual results differed from what they predicted (higher percentage for women, lower percentage for men). That is the crux of the study.

I’m not the only one who picked up on this travesty of twisted statistics. More than 1100 comments follow the article, many taking the writer to task for manipulating the facts.

Please, fellow writers. Do NOT try this at home.


No comments:

Post a Comment