Dentists named Dennis, Georgias who move to Georgia, free will has nothing to do with it, confusion about conditional probabilities

John Shonder points me to this article on the work of Brett Pelham, who’s been featured here before. The news article states,

In studies involving Internet telephone directories, Social Security death index records and clinical experiments, Brett Pelham, a social psychologist, and colleagues have found in the past six years that Johnsons are more likely to wed Johnsons, women named Virginia are more likely to live in (and move to) Virginia, and people whose surname is Lane tend to have addresses that include the word “lane,” not “street.”

They didn’t mention my favorite, which is that there are almost twice as many dentists named Dennis in the United States, compared to what you would expect based on the number of dentists and Dennises alone.

Nooooooooooo……………

I want to correct one misconception that was aired in the Times article. As with many such things, it turns on conditional probability. The article states,

In studies that make believers in free will squirm, Dr. Pelham’s team asserts that names and the letters in them are surprisingly influential in people’s lives. . . . Skeptics of the name-letter effect question how strong the affinity really is between a person’s name and his or her destiny. ‘I’m willing to believe that such patterns exist,’ said Stanton Wortham, a professor of education and anthropology at the University of Pennsylvania. ‘But I’m not willing to grant that those sorts of patterns are going to explain or drive a substantial amount of behavior.'”

OK, first off, free will has nothing to do with it. Everybody agrees that your party identification and, for that matter, your religious affiliation, are highly correlated with your parents’; does this mean you don’t have a chance to alter these things? Free will requires the ability to alter things; it doesn’t require complete statistical independence of preconditions and outcomes. Just a moment’s thought, blah blah blah.

On to the second point. The pattern of names and occupations (for example) can be clear and still represent a small effect. Just for example, there were 482 dentists in the United States named Dennis, as compared to only 260 that would be expected simply from the frequencies of Dennises and dentists in the population. On the other hand, the 222 “extra” Dennis dentists are only a very small fraction of the 620,000 Dennises in the country; this name pattern thus is striking but represents a small total effect. Some quick calculations suggest that approximately 1% of Americans’ career choices are influenced by the sound of their first name.

P.S. I agree that stories about names are amusing.

4 thoughts on “Dentists named Dennis, Georgias who move to Georgia, free will has nothing to do with it, confusion about conditional probabilities

  1. Well, isn't this outcome consistent, perhaps unlikely but consistent, with people named Dennis choosing occupation at random? perhaps unlikely but possible. How can one distinguish between a real effect and a random event in a topic like this?

    C

  2. This sounds like an alpha version of Kahneman and Tversky's anchoring effect — if you hear another number out of context, say "15" or "77" and then somebody asks you what the price or tuna is, you are more likely to give a number like 15 if you heard 15 before, or 77 if you heard 77 before.

    Another close relative would be Bob Zajonc's effect of mere exposure, which says that the more you are exposed to something the better you like it (ceteris paribus). If you are named Virginia or Dennis, it's not surprising that you'd find these syllables more harmonious.

    I wonder: people who go to Michigan like the color blue after they attend school there. Did they like blue more years before? Somebody's probably studied that.

  3. C,

    The numbers are too large for the pattern to have occurred at random.

    Z,

    I don't know about Michigan, but I think the authors of those papers considered some of these models.

  4. One of the issues I have with the "Dentists Named Dennis" study, is that it assumes that name distribution is equal across a population. That is, if 4% of the population is named Dennis, then 4% of the workforce is named Dennis. Given the fluctuation of name probability over time, this is wildly inaccurate. In fact, while Dennis was the 20th most popular male name in the 1940s and 50s, and remained relatively popular into the 1960s, it's now something like the 330th most popular name. So for men in the workforce (i.e., eligible dentists), Dennis is a much more common name than it is in the population at large. I could be wrong, but I don't think the study actually accounted for this issue.

Comments are closed.