- ‹ previous
- 2009 of 2798
- next ›
And Now For Something Completely Different
AND NOW FOR SOMETHING COMPLETELY DIFFERENT....Yesterday I posted a chart that plotted the frequency of prayer vs. partisan affiliation. It came from Razib Khan, who created it using data from the University of Chicago's General Social Survey. Today, although he was happy that a few blogs linked to it, he lamented that "alas, the practice of looking to the GSS to test some intuition or CW hasn't spread like wildfire."
Well. That's like catnip around here. I myself have never done this for a simple reason: I didn't know I could. But it turns out that some fine folks at Berkeley have built a simple web interface for the GSS and several other big databases (here), and anyone who feels like poking around can do so. So I did.
The interface lists all the questions that the GSS asks and allows you to plot variables against each other to see what pops out. I did that for a while, generating nothing of any value, until I finally discovered something of vital importance: one of the questions on the 2006 GSS was, "How many people named Kevin are you acquainted with?" And there was even a followup question: "How many of those people named Kevin do you trust?"
I had to find out. Now, I could have plotted this against anything I wanted — age, sex, religious attendance, zodiac sign (really) — but this is a political blog, so I plotted it against party affiliation. The results are on the right, and they're a little disturbing. Eyeballing the numbers, people appear to know an average of two Kevins each, but they only trust about half a Kevin each. So on average, people only trust about 25% of all the Kevins they know.
That's a little deflating, isn't it? But interesting! If you're named Kevin, that is. You can check out results for your own name, but only if your name is Kevin, Karen, Shawn, Brenda, Keith, Rachel, Mark, Linda, Jose, or Maria. Your guess is as good as mine about why they chose those ten.
You can also do other stuff, of course, and that includes mining the data and abusing the results to produce results you find pleasing. And then blogging about it. You can probably expect some of that in the future. Until then, have fun.
Say Kev-here's a chart for you to plot-Crime rate in the U.S. vs Supreme Court decisions on Separation of Church and State. OR, Disasters vs taking God out of the Pledge and off of coins. Or, Corruption of our National Leadership and Church Attendence-They could prove interesting!
im in the 25 per cent
I question your methodology. What is the definition of "acquainted" with a Kevin?
For instance I have probably been acquainted with about 3 1/2 Kevins in my life (Drum is the 1/2) but the only one that I consider a friend was when I was growing up and I haven't seen him in almost 20 years. Am I still acquainted with him? As a kid I would have trusted him but now probably not. The other was a co-worker and another a classmate. The co-worker I definitely don't trust and it depends on what I am trusting the classmate with.
So I think your methodology is flawed.
The fact that "Keith" and "Rachel" are on the list of tested names is surely of great interest to MSNBC.
The opposite of "I trust Kevin" is not "I distrust Kevin". There is a third possibility -- "I don't know Kevin well enough to know if I can trust him".
In view of this, 25% seems like a pretty good score.
The interface lists all the questions that the GSS asks and allows you to plot variables against each other to see what pops out. I did that for a while, generating nothing of any value, until I finally discovered something of interest:
You'd make a great social scientist - you have an intuitive understanding of data mining. I see a great new career for you.
Clearly they didn't poll any cats on this Kevin question.
As a Kevin I find this disturbing. Unsurprising because we're a shady lot but disturbing indeed.
No doubt they were fishing for what people think of Jose and Maria and did not want to seem too obvious.
Nat, LOL. Seriously, for those who want to know, I believe that these questions are intended to discover how many acquaintances the average person has (and how many he trusts). That is, given how many Kevins you know, and an estimate (from birth certificate data) of how many Kevins there are in the general population, I can deduce how many people you know.
But the infinite creativity of Mr. Drum will find even better ways to use this data, I am sure.
I, too, only trust half a Kevin.