Thursday, January 14, 2010

Must be important; it's on Facebook

One of my readers, a former University of Maryland at Baltimore County student, noticed a couple of articles at that cited Facebook numbers. His comments:

Both stories cite the membership of a Facebook group related to the story as a substantial fact (the first in the seventh paragraph, the second in the 18th. Both figures are between 1,000 and 2,000, but neither provide any context such as how quickly that figure was reached or even e.g. how many people there are in total on Facebook in the County or the City, or what constitutes a significant number for a group of this sort.

It wasn't so long ago that I was a college student, and I don't recall the joining of a Facebook group as a particularly meaningful act (compared to, say, writing a letter or attending a demonstration). I'm sure checking Facebook is now an early step in the first pass on reporting any story, but it doesn't seem to me either of these numbers _mean anything_. Would this give you pause as a copy editor?

Oh yes. Nearly anything will stimulate suspicion in a copy editor, especially an article that cites numbers.

Nothing is easier than clicking on a link to join a Facebook group. I have joined a few, and some I have not gone back to look at in months. I can see raw numbers of membership in these groups, but, apart from comments by a fraction of the members, I have no idea how often or how much anyone participates in them.

What, as a copy editor, I suspect, is that these citations of Facebook numbers are analogous to those bogus Internet polls that Web sites love to set up and then quote.* Any poll, the ready reader understands, that depends on self-selected responses is statistically invalid. Dubious. Suspect. Worthless.

Trying to gauge involvement with an issue by a raw count of Facebook members in a group, without the availability of a metric to measure the degree of involvement, is just one more largely meaningless number.

*You may have seen some of these ludicrous polls that show the percentages, for and against, in large type, and then reveal in the small type beneath that the poll drew no more than two or three dozen responses.