The Washington Post reports today on a Harvard study that claims “‘binge’ drinking by college students was significantly lower on campuses with more female and more black, Asian and other minority underclassmen.”
Previous studies have shown that younger, white male students, particularly fraternity members, are at higher risk for binge drinking, which is defined as having five drinks in succession for men and four for women. The Harvard study does not conclude that the presence of minority students directly results in less binge drinking, but rather that there is a correlation between diversity and less heavy drinking by students.
For a serious and thoughtful critique of this study (at least as reported), see Erin O’Connor’s comments today. I’m afraid I, however, can’t take it seriously.
That’s not to say that I think it’s wrong, its methods flawed, its conclusions suspect. They may all be right as rain. It’s just that the leap from “correlation” to causation seems to be more one of faith than science. (Again, Erin makes this point much better than I, and she manages to do so without cracking up, which I can’t.)
Meanwhile, until more data are produced, and assuming that the correlation really does exist, here are some possible partial explanations that come to mind:
1. Binge drinkers tend to be white boys. The more girls in the student body, the fewer boys. Ergo….
2. Binge drinkers tend to be white boys. The more minorities there are in the student body, the fewer white boys. Ergo….
3. Binge drinkers tend to be frat boys. Women don’t join fraternities. More women=fewer frat boys. Ergo….
4. Binge drinkers tend to be frat boys. Minority students are less likely to join fraternities, resulting in fewer frat boys. Ergo….
5. Booze costs money. White/frat boys have more disposable income than minorities. The more minorities, the fewer white/frat boys. Ergo….
But then, whoever said “diversity” wasn’t a sobering experience?
Regarding the Harvard binge drinking
study — maybe their next step is to
determine how many of the binge
drinkers had Irish surnames?
How would this be any more offensive
than the obvious stereotyping of
“white males.”
That’s not to say that I think it’s wrong, its methods flawed, its conclusions suspect.
Obviously, you don’t believe that and neither do I. These studies give junk science a bad name. First of all, studies such as these are inherently flawed. There is no way of validating the information. How do you get reliable data on binge drinking or sexual activity etc.? You must ask the people and accept their answers as true. That is a big stretch.
The study found that on campuses that had 21 percent or more minority representation, 44 percent of students were heavy drinkers. That number rose to 53 percent on campuses with less diversity.
If doing bad science were a crime, these people would get the death penalty! If you were trying to show a relationship, then at a minimum you would show that there was a direct correlation between the percentage of minority students and binge drinking. In other words, you would use more that TWO points in your graph. I would bet anything that if you plotted percentage of minority students against the percentage of binge drinkers you would not get a curve that was monotonically increasing. By lumping the data together, they conveniently hide this information. Also, why use 21%? What is magic about that number? Do you think when you increase the number from 20% to 21% suddenly all them White boys say I don
Before you guys get too bent out of shape, make sure you look at the study itself. Sometimes (OK, frequently) news reporters with deadlines don’t really go very deeply into a study; usually the study itself is pretty nuanced and guarded, and is harder to find fault with. (For an example, see my posts on the Atlantic Monthly’s extraordinarily poor reporting on a very interesting study about income differentials between hetero- and homosexual workers (here and here)).
That’s not to say the newspaper’s analysis isn’t a legitimate subject of attack, but I think it’s a mistake — at least if you’re interested in truth and not just polemic — to attack a study only on the basis of a newspaper article about it, since, unfortunately, newspaper reporters don’t always have the firmest grasp of statistical terminology or the time to do much more than read and misunderstand the abstract of a 50-page study, and then call people and see what they think about the reporter’s interpretation of the study’s finding. That may or may not be the case here, but I hope you follow up on this when the actual study becomes available, if it isn’t already.
Garrett,
What part of “this study is junk science” don’t you understand? These types of studies are inherently flawed. You need valid data, the results must be statistically significant AND you must be able to show causality to be able to draw any conclusions from the results of the study. Since this study lacked all three of these, what else makes you think we should not reject it as propaganda for a particular political point of view?
Garrett
“… unfortunately, newspaper reporters don’t always have the firmest grasp of statistical terminology or the time to do much more than read and misunderstand the abstract of a 50-page study, and then call people and see what they think about the reporter’s interpretation of the study’s finding.”
That’s more than some do. Distressingly common is where the reporter simply rewrites the press release announcing the publication of the study. After all, many press releases now include quotes from the author, background about the field, and judgments (usually inflated) of the study’s significance.
Recently there was an article out of UCLA with some tentative indication that gender begins to be set before there are any functioning gonads or sex hormones directing development in the fetus (which may help explain why “gender identity” isn’t a simple thing, and may help to deal with the surprisingly large number of people who are born with both male and female sexual equipment). A Reuters story butchered the press release but got picked up by a large number of news outlets.
Without seeing the list of the colleges that were studied, its hard to tell, but I’m guessing a major confounding variable is eliteness of the school. Harvard, Yale, et al select for a class low in students likely to binge drink and high in minorities, which would cause most of the reported results right there. Shoddy, shoddy work.
Patrick McKenzie
I’d wager dollars to donuts that binge drinking would also be lower at institutions that:
have comparatively high numbers of evangelical/fundamentalist Christians
maintain high enrollment of Mormon students
enroll significantly large numbers of orthodox Jews
promote their religion departments, even at the expense of narrow “_____ studies” programs
have larger proportions of faculty (as opposed to T/As) who teach classes
make classroom teaching a meaningful part of the tenure process
require professors to teach morning classes on Fridays
regularly schedule Saturday exams
aggressively use the police to root out underage drinkers at off campus venues like houses, apartments and bars and not just fraternity houses and dorms
Anyone want to bet whether the Harvard study authoers will take me up on any of these conjectures…or whether the indolent professoriate would support any of them if the research bore them out?
John,
You missed the line in the intro to the study, drinking among individuals in high risk groups (ie:white boys) was lower at schools that had more diversity. For example (caveat: this is a hypothetical not based on actual numbers) fifty percent of high risk males on non-diverse campus engage in binge drinking but only ten percent of high risk males on diverse campuses engage in binge drinking. Therefore most of your explanations do not work.
Stu Gittleman raises some valid points, although hard core enforcement tends to increase binge drinking. The more underground the drinking, the more it takes on a binge quality.
As for whether this is junk science or not, I’m an oral historian by training, I’m pretty comfortable with self-reporting and qualitative data by training if it is used critically. My understanding about alcohol studies is that participants tend to under-report consumption if they are still heavy users.
Patrick McKenzie is flat out wrong on the question of binge drinking at elite universities. These tend to be places with high binge drinking rates (MIT for example). However, because lower echelon state schools tend to be filled with students who are also working close to or completely full time, or have kids, or both, those students have much less time to drink, much less binge drink period.
Finally on the question of whether you would get a curve or not. Tipping points happen all the time in human behavior. I would not expect to see a gradual curve, but a tipping point as more social options open up.
One possible reason is that at your typical non-denominational college your white-bread suburban kids grow up on a culture of keggers. This tends to continue in college. Nobody is offering them any other options. If they go to a school where people from a lot of backgrounds who grew up not depending on alchohol (whether it is religious, economic, social, or locational reasons) those folks will have more ideas and create more options. Do I have statistics to back this up? No just qualitative experience teaching at 2 colleges and one high school.
The anonymous commenter above may not be “flat out wrong,” but he’s close:
“However, because lower echelon state schools tend to be filled with students who are also working close to or completely full time, or have kids, or both, those students have much less time to drink, much less binge drink period.”
I can safely say that you’ve never been to a state school, then. Look at the Princeton Review’s list of top party schools: almost exclusively state schools. You’re forgetting two things: one, sports increases drinking (tailgating, parties, etc), and two, beer is cheap.
Further, even students who work full time (40 hours) and go to class full time (18 hours max, usually 15), are only spending 55 hours a week at their work & study. 7 days times 8 hours of sleep a night is 56 hours of sleep (and what college student really gets 8 hours of sleep a night)? That’s 111 hours a week sleeping, working, and studying. There are 168 hours in a week.
So, with 57 hours of free time – just over eight hours a day on average – you want to tell me that these kids don’t have time to drink? Tell that to the hundred or so people who emptied the kegs at the party I was at on Halloween, heh.
Hm, I did not mean my last post to be anonymous. I thought “forget personal info” had to do with whether or not my e-mail would go into some database. Anyway, if I recollect the list of top party schools it included schools like UC Boulder (the top state school in Colorado) and Michigan State (1st or @nd school in Michigan depending) it did not include places like Adams State (in Co.) or Wayne State in Michigan. Plus your count forgot study time, commuting time, etc. etc. etc. BTW I was teaching at a top state school for 10 years, I know how much drinking goes on there. I lived 2 blocks from the stadium. Sad to say, the people who peed in my yard and threw up on my sidewalk were well past college age.