Nametrix: Doctor or Dancer?

January 24, 2013 11:13 PM

Nametrix: Doctor or Dancer?
Nametrix is an iOS app I created initially for baby naming, but it ended up being way cooler than that. It uses public data on millions of people (US census, US campaign contributions, Wikipedia) to determine what professions, political parties, and other affiliations each name tends toward. For example, Ellen is a disproportionately common name for 1) psychotherapists, 2) librarians, and 3) activists. Ellens also overwhelmingly lean toward the Democrat party and have tended to be most popular in the northeastern part of the US. As far as I know, nobody else has made anything remotely like this... Pretty fascinating stuff, right? What other kinds of metrics would you like to see?

Role: developer

posted by hodgebodge (18 comments total)

Looks like you opted for the (unnecessarily restrictive IMO) iOS6 compatibility. Shame that, I would have enjoyed this from time to time if it worked on my device :(
posted by a_green_man at 9:41 PM on January 25

Sorry, a_green_man, I hadn't originally intended to restrict it to iOS6+, but I discovered the hard way that I'd used some iOS6-only features. With the iPad version, I think iOS6 really improves the experience, in any case, since it enables the app to have a nice cover-flow style navigation through the pages of metrics. Hope you get to enjoy it when you eventually are able to upgrade to iOS6!
posted by hodgebodge at 12:28 PM on January 26

By the way, if you guys like it or hate it, please take just a few seconds to give it a star rating in the app store. It'll give the app more exposure and encourage me to spend more time on stuff like this. :)
posted by hodgebodge at 12:43 PM on January 26

This is really extremely cool, and very easy to use. It has a lot of extra features like graphs of popularity over time. It's the kind of thing you could make some instant friends with at a social gathering. Truly great work. Good luck with it!
posted by Miko at 8:02 PM on January 26

One thing though, it seems hard to believe how often careers like "casting director," "model," and "art director" come up. It seems much more common than those careers are in real life. Is that because the database you're pulling careers from has something to do with media mentions? I know a lot of Sheilas, and none of them are casting directors, though that's the #1 profession appearing under that name. Very few people are actually casting directors.
posted by Miko at 8:04 PM on January 26

Miko: What's going on there is that the data includes a lot of people in those professions since around half a million of the names in the data set are from Wikipedia, and Wikipedia of course skews toward people with some sort of prominence. So, a lot of first names will be represented in those professions (e.g. casting director), and those professions will be near the top for those names that happen to be disproportionately common in those professions.

A key here is "disproportionately common"... The highest total number (200) of Sheilas might be businesswomen, let's say, but if 5% (20) of all 400 astronauts were named Sheila, and only 2% were businesswomen, astronaut will rank higher for that name. There are minimums, by the way, so several instances of a Sheila in a profession with only a small total number of people won't be included (not statistically significant enough).

This is why "John," the most common name, isn't the top name for every profession. We're looking at percentages of the Johns in each profession, not the total counts (pretty meaningless).
posted by hodgebodge at 9:01 PM on January 26

As you can see, it's tough to describe the fact that the app uses relative frequencies rather than absolute counts, and I'm a bit worried that some people will think that something's broken and give the app a low rating. It's a real challenge to get that across in the app itself without cluttering it, but it's something I continue to work on.
posted by hodgebodge at 9:35 PM on January 26

We're looking at percentages of the Johns in each profession, not the total counts (pretty meaningless).

Yeah, I wish it could give an actual breakdown of how people's names are distributed across professions, rather than the professions that have an unusually high number of X names. It is interesting but, as you say, it really skews, and it was easy for me as a user to misunderstand what was being implied. Part of me thinks it would be more interesting to see what professions all Sheilas most often have, rather than what professions have the most Sheilas. For baby naming that's almost more important.

Still, it's really cool.
posted by Miko at 6:18 AM on January 27

The intent isn't to show what profession the highest number of Sheilas are in -- that's likely to be the same profession as for every other name. You'd see more or less the same list of professions in the same order, regardless of the name, just by virtue of some professions being far more common than others.

Instead, the app shows that Sheilas are overrepresented in the casting director profession, more so than in any other profession for which we have significant data. Thus, there's something about the casting director profession that attracts Sheilas more than any of the others, once you've normalized the data to reflect the fact that not many people can be casting directors.

I agree that it would be cool to have counts of each profession for each name, but without the normalization, it's going to look really similar for each name. The app essentially brings out the anomalies that show how names are actually subtly different.
posted by hodgebodge at 9:35 AM on January 27

ok.
posted by Miko at 10:10 AM on January 27

Cool. Thanks for the questions and encouragement. Definitely lets me know that making the app's statistical undepinnings more straightforward more of a priority!
posted by hodgebodge at 10:58 AM on January 27

Aw. This sounds neat. Let me know when you have an Android or Web version!
posted by limeonaire at 3:13 PM on January 27

Will do... Android's next on the agenda.
posted by hodgebodge at 9:27 AM on January 28 [2 favorites]

This really is a gorgeous app. Kudos!

Looking through the top lists of names by profession is very funny. You get such wonderfully stereotypical names. I've been having fun presenting my girlfriend with a list of names and seeing if she can guess the occupation. e.g.:

a) Landon, Chaz, Brock, Tristan, Cole...

b) Freddie, Rocky, Benny...

c) Ivana, Brandi, Eugenia, Juliana, Kylie...

d) Dominick, Eldon, Delbert, Joaquin, Nolan, Doyle...

(Answers ROT-13ed: KKK cresbezref, obkref, zbqryf, rkrphgvirf)

It could be fun to build this in as a multiple choice game.
posted by painquale at 8:15 PM on January 29

Love it. Pony request: shake the app, and get taken to a random name.
posted by roger ackroyd at 11:02 PM on January 29 [2 favorites]

Pretty big news: the app was on Good Morning America yesterday. Apparently, George Stephanopoulos looked himself up... Archeologist is one of his name's top professions.
posted by hodgebodge at 9:22 AM on January 31 [2 favorites]

Another fun experiment might be to link the names, also, to thinks other than professions. For example, which names are most likely to be involved in a sex scandal, become alcoholics or reach the Forbes top 500 most wealthiest, etc. ;)
posted by misha at 9:43 AM on February 11

Yeah, I've done that with profession, political party, and geographical region of the US so far and would definitely like to do a lot more. One of the biggest challenges is getting huge (for real statistical significance) data sets with first names included. So, sex scandal might not be doable, but I like that line of thinking. :)
posted by hodgebodge at 9:50 PM on February 11 [1 favorite]