Princeton Class of ‘77 Loves Their Kids, But Not as Much as ‘82
It's summertime, the grades are in, and the students are all gone--what's a provost to do but count up the change in the piggy bank? Princeton has been doing just that, and reports that this year they've received a staggering record of $49.04 million in alumni donations, a 21% spike from last year.
Reasons? "'Frederick G. Strobel '74, the volunteer leader for Princeton's annual giving campaign for the past three years, said alumni donate to Princeton so readily in large part because of the university's "commitment to excellence.'"
And I got an *awesome* Livestrong bracelet because I actually care about Lance Armstrong's testicles.
A better explanation can be found in a research document on Princeton's own web server: Altruism and the Child-Cycle of Alumni Donation (pdf). Using data from an "anonymous selective research university," Stanford's Jonathan Meer and Princeton's Harvey S. Rosen toss about logarithms, Greek letters, and Tobit estimators to conclude that "about 52 percent of giving by parents whose children apply to Anon U is due to altruism and the remaining 48 percent is due to self-interest."
Maybe the best explanation for increased Princeton giving is the increased numbers of pimply teenagers applying, which in turn is probably due to increased unprotected Princeton alumni sex back in 1988-1990. Don't believe us? Princeton giving broken down by class year. Graduate in '82, kids by age 30, progeny aiming for class of 2012 equals $8 million, or a sick average of $10,267 per alum... probably taped to the back of Jr.'s application. And they said dropping Early Decision would hurt the bottom line!
--BETH MILTON
17 Comments | 
| 
Print This Post
Read more: alumni, guest editors, Princeton
Email – 
Search
About
Follow us on Twitter
Report a bug
Archives
RSS Feed
July 17th, 2007 at 3:48 pm
i thought, damn another post about princeton. but this is actually not about princeton but rather some damn good sociology.
July 17th, 2007 at 3:51 pm
“Princeton” is a single entity. It is not a “they.”
July 17th, 2007 at 3:56 pm
Slate ran an article on this trend about a week ago. http://www.slate.com/id/2169858/fr/flyout
July 17th, 2007 at 4:13 pm
It was their 25th reunion… they got drunk and started writing cheques.
July 17th, 2007 at 8:53 pm
“And I got an *awesome* Livestrong bracelet because I actually care about Lance Armstrong’s testicles.”
I laughed out loud, touche.
July 17th, 2007 at 10:13 pm
I guess my mom ought to have waited to coordinate with the rest of the ’82s before having unprotected sex – she would have known to send a bigger check with her alumni mag subscription. And I always thought I didn’t get in cuz I ain’t read no good.
July 17th, 2007 at 10:51 pm
http://www.metaezra.com/archive/2007/06/why_legacies_are_given_prefere.shtml
July 18th, 2007 at 12:04 am
Where is this so called Princetown anyway?
July 18th, 2007 at 12:38 am
So whoever posted the ‘MetaEzra’ comment was not associated with the actual blog. We aren’t ones to engage in blatant self-promotion.
July 18th, 2007 at 1:10 am
Don’t get so excited here. I can’t be 100% certain from reading their paper; but, it seems very likely that they’ve been sloppy with their data. The technicals: They’ve likely got correlated data — and didn’t introduce random effects to model the correlation structure and give realistic standard errors for their model coefficients. I mean, COME ON. Look in the appendix. A model with tens and hundreds of significant terms? No. Unh unh. Bullshit.
The effects they mention are true. (I’ve seen proper analyses of data from MIT.) But, their faulty analysis precludes any guess as to proper magnitudes of the effects, standard errors, or (ergo) significance.
This is the travesty that occurs when sociologists in the guise of economists try to do statistics. Somebody should take away their PhDs and spank them.
July 18th, 2007 at 10:34 am
The standard errors are clustered based on individuals to control for correlation, and there are roughly 500,000 observations in their data set — more than enough variance to include ~100 variables and still obtain meaningful standard errors.
But let the great, ever-lasting debate between econometricians and statisticians begin.
July 18th, 2007 at 9:38 pm
Clustering based on individual is not sufficient. It’s still just using fixed effects — so the s.e.s they calculate will be artificially small. Additionally, they’ve neglected effects due to autocorrelations and cross-correlations among classes and regions.
Sure, they’ve accounted for all of these with some fixed effects. But that’s the joke: it is not good enough. Fisher saw so over fifty years ago. Check out a text on longitudinal data analysis (I recommend Diggle, Heagerty, et. al.) to see what havoc this can cause.
Like I said: the effects are there; but, they’ve bumbled the analysis. Further, they did this crazy semi-non-parametric approach that makes interpreting the model way too difficult. A basic intercept, time slope, time quadratic (to allow for some nonlinearity), and categorical variables like they have — plus interactions — would have yielded a much smaller model that would have been easier to interpret.
And it gets worse. Go look in the appendix. Why do they have a parameter for every year of a child’s age? Of course those are likely to all be non-zero; that’s irrelevant. Kinda like measuring salaries of workers with a GED, BS, MBA, and PhD — and noting that all the average salaries are significantly different from zero. Who taught them to present results in such a manner; it is misleading at best and fraudulent at worst.
Econometricians and statisticians get along in all my experience. It’s the economists who hide behind a wall of poorly-done statistical analysis who raise hackles.
July 19th, 2007 at 2:06 am
Stats said: “This is the travesty that occurs when sociologists in the guise of economists try to do statistics. Somebody should take away their PhDs and spank them.”
I’m sure you have a point regarding the statistics, but you’re dissing Harvey Rosen, who is regarded to be one the foremost microeconomists in the country right now… he was head of the white house economics council for a while, too.
July 19th, 2007 at 5:02 pm
Yes, I know who he is. And yes, I think his methodology was flawed. I don’t believe in “proof by reputation”. Using random effects is advanced undergrad/masters-level stats.
Compare Rosen’s paper to Levitt’s latest (on school choice) in Econometrica. Levitt keeps things simple. That lets him see intriguing features in the data and lets him tease much more out of it.
Am I being too harsh? No. Look at the history of excellent economists and many of them were also faculty-quality statisticians: Engel, Edgeworth, Hotelling, Friedman, Stigler, Bollerslev, Engle, Granger, Heckman, Levitt, Hansen, and so on.
July 21st, 2007 at 10:41 am
the first article I saw about this was this: http://www.insidehighered.com/news/2007/06/14/alums
August 15th, 2007 at 1:20 pm
Stats might have been worth taking seriously, until he betrayed his ignorance by requesting “random effects.” Random effects require unpleasant orthogonality assumptions that are invalid in studies like this. More to the point, had he (or she) actually _read_ the paper rather than doing the classic undergrad reading technique of flipping to the tables of results, he (or she) would have found that most of his (or her) comments are addressed.
August 15th, 2007 at 1:23 pm
I’m starting to think that “Stats” is an overexcited undergrad majoring in statistics. Two more dumb assertions he (or she) makes: unlike the GED/BS/MBA/PhD example, this paper measures the effects against a comparison group. Finding significant effects there is actually meaningful. Also, there’s nothing crazy, non- or even semi-parametric about this approach. “Stats” probably read those words in his (or hers) shiny advanced statistics textbook and really wanted to use them.