Regression to the mean

After starting the season 0-6, the Boston Red Sox are now seven games over .500 and in first place within the American League East.

The middle

To the relief of Boston fans and the chagrin of Yankees fans, the statistics principle of regression to the mean comes into play. Boston wasn’t going to lose them all this season; if you look at the team’s history since 1901, the team has a lifetime .515 record. As you can imagine, there are good years and there are bad years. The worst was .279 in 1932. The best was .691 in 1912.

I see this principle playing out in social business, particularly in the number of services individuals use. We all started out using few, if any, “social” tools. Email and static web sites welcomed blogs, wikis, and social networks. Some of us experiment with many services focusing further on identity, location, and distributed publishing. But ultimately, individuals come back to using a core set of tools.

I get asked to try out a lot of social tools, both internal and external. Eventually, I always regress to the mean and find myself using a handful of core tools. New offerings can replace old ones, e.g. Mindtouch + Socialcast instead of a static intranet, but the total number in use moves up slowly, if at all.

Baseball looks pretty similar from year to year, but shows greater difference in decade to decade. The differences are only apparent when stepping back and looking at a broad time frame.

Leave a comment

Comment now or forever hold your peas

This site uses Akismet to reduce spam. Learn how your comment data is processed.