I really want to read the thesis but can’t get anything but thumbnails and it is too difficult for me to read it that way. Suggestions?
Here are few observations from a former sorority girl. Sorority girls don’t need to go to fraternity houses for free alcohol. They will have no problem acquiring fake IDs. They can probably walk into any bar in town and men will happily buy them drinks. Some of those men will be delighted to come pick them up at sorority houses the next weekend and take them to bars or clubs to drink and dance. A group of sorority girls have the ability to make a party happen wherever they show up. They really aren’t at those fraternity parties for the alcohol. They are there for the high status males. It is about access to power. It used to be about acquiring some of that power through association, especially through marriage. I am going to be surprised if that isn’t still the case. Sorority girls, who want to have sex with fraternity boys, can have sex someplace other than the fraternity house. In my day, this is what high status sorority girls did. They had enough power to dictate where they would have sex and it wouldn’t be at a fraternity house where one had to be concerned about the punch. They were not going to be walking home alone the next morning.
Sorority girls sometimes (always!!??) like sex as much as fraternity boys. College is a good time to experiment. The double standard (which used to exist and I believe still does) means the higher a girl’s status, the less negative impact multiple sex partners will have on her later life. It won’t negatively impact her acquiring power through association with high status, powerful males. If that double standard no longer exists, and all girls want sex equally, the higher status girls may just be more successful achieving their goals.
Someplace, a few months back, I read an article reporting results on research comparing low and high SES college women and slut shaming. The article said low SES women were slut shamed for fewer sex partners than high SES women. I hope I am reporting that correctly and not misremembering. Does anyone else remember reading that? I am pretty sure someone linked to it on one of these threads.
Just like some find the bad boys attractive, some find the bad girls attractive. At least in my world there was the stereotype of the wild rich girl, tearing around town in a fancy vehicle her parents bought her, and flaunting all the rules. One of my much younger cousins had a pink jeep. Her mama asked me about the daughter of a famous politician, “Have you met her? Is she really wild? I hear she is just tearing it up at college.” There was a look of admiration on that mama’s face. My cousin eventually married a powerful man and become a pillar of the local society in which she grew up.
High status fraternity boys don’t have to rape girls to have sex partners. They don’t have to get them drunk or even provide them alcohol. Girls are more than willing to have sex with them sober. So I don’t think this is about alcohol.
I do think college kids believe alcohol excuses actions they wouldn’t take sober Unfortunately, somehow we modeled that behavior for them. If they think they need alcohol as an excuse to get busy, they may think what they are doing is wrong and that is just so sad. I started on these threads believing drunken college debauchery was not necessarily bad and maybe just a part of growing up. I moved beyond that belief several threads back.
Thank-you. I was also wondering about the premise which struck me as odd that sex should be easier to ration by sorority girls because they have higher status in some but clearly not all circles because it was certainly not my impression that they necessarily had any interest in rationing anything. Sex and rape don’t really have all that much in common until you get to the grey area cases which is where I do think alcohol plays a very large role for both men and women. However, that post is very helpful.
alh: If you click on the thumbnail and then look up at the top, a PDF is created and if you click on that, open a readable pdf of the page in another tab. Kind of annoying. I couldn’t figure out a way to open the entire PDF. I tried reading it, but ended up skipping much of the statistical analysis. My use of statistics is limited to more straight-forward environmental dat so some of the sociological statistical analysis was not easy to follow. I skimmed through much of it, reading the first part about the survey results. Conclusions start on page 58.
Interestingly, he did not get much of a response from fraternity guys so all of his analysis is based on the responses from women. I also thought it was interesting, and probably makes sense, that women that have higher social status to start with can have more “hook ups” without losing status. Lower status women were labeled as slutty much more quickly.
The article, particularly the comments from the past sorority presidents, was more interesting to me. The comments on the article are not that helpful, although one said there are actually 60 organizations. Not sure the point of that is, except to say that maybe the methods were flawed.
As someone said upthread, looking at this issue as parents and what we remember from our own college days may not reflect what is really going on today. I was not at all into the Greek scene and was not at a big school with dominant greek life. I remember girls may have participated in very occasional one night stands, or had relationships. Those relationships may have been very short-term, but they were relationships. It did look like from the thesis that the number of hookups per woman was higher than the number of partners so it may be that some of these were with the same guy. Or that could have been the way the data was presented.
The thesis did not really discuss rape vs hookups.
On the right half of the screen, a couple lines down from the title, is a button marked “Save”. Click on that, then “Download”. You can then save the pdf to your computer / open it directly.
Thanks for sharing your thoughts. I have some too, but it doesn’t look like I’ll have time to write much today.
A few points for readers though - the thesis says what it says. But the article’s author (Nathanial Hess) is an editorialist who makes a lot of claims, some of which have some research backing but a lot of which seem to be just this guy’s opinions.
For example, the thesis deals almost entirely with consensual sex. The leap to making claims about sexual assault is done by the editorialist. So without reading the thesis I’m not sure how much of the article’s claims are just some 20 year old’s agenda with the thesis merely serving as a tool to give them legitimacy (not to mention that lots of social science research is as reliable as homeopathic medicine).
Interesting thesis. I have a feeling that the author is trying to get more out of the data than is in the data, and the entire business of a quasi-Poisson quasi-probability distribution looks bogus to me.
However, lots to see. Most women, both sorority women and independent women, report not hooking up at all. The women in the “top” sorority houses drink more than other women, hook up more, have more different hookup partners, are thinner, and say that they are more attractive.
FYI - You’re right that the statistical analyses and the various regressions don’t shed very much light on the data beyond what’s readily apparent.
But that’s ok - remember it’s just a master’s thesis by a student, and his advisor is teaching him. The student is just showing that he can apply some of the basic technical tools from his econometrics class to his thesis problem.
All the insight is in coming up with a very interesting question, designing the survey, trying to figure out how to get a bunch of fraternity and sorority kids to even answer the survey somewhat honestly (and cleaning the dogsh*t data that I’m sure came back). Once you have clean data then analyzing it is simple, but that’s to be expected.
I think part of the problem on college campuses today is that being in a steady relationship no longer has any status whatsoever. In the early 80’s having a boyfriend/girlfriend still had some status. And I think those relationships, even the ones that only lasted 6 to 9 months, kept us away from some of the issues women now face. Relationships were more common and sexual needs were being met in a healthier manner, rather than this “nightly hunt” for a partner that appears to be the norm today.
Attending a party with your boyfriend or a group of couples was a very different experience than attending with a group of your girlfriends. Looking back now, I think the guys knew what the girls would be up against at those parties, because they certainly didn’t like you going to certain parties with just your girlfriends. Today, it seems that being in a relationship is considered outdated or passé - for high school as well as college.
@al2simon, do you understand the quasi-Poisson statistics? I can’t understand what the underlying model is. I know what a Poisson model is, I know what a negative binomial is, but what is a quasi-Poisson likelihood in real life?
If you read my post, you would note that I was very specific in what I said, and you give an argument in the general. I was never talking general. And schools do have speech codes where certain words are considered inappropriate to use in certain places and times, no different then not yelling fire in a movie theater.
That said, I have zero issues with freedom of expression etc., and Emma has the privilege to do her project about sexual assault if she likes.
However, her project is not just a freedom of expression; it is a targeted project which underlying aim is to attack one person and get him off campus, even though he was found not responsible for harming her as she charged. In fact, she said she would not stop until Paul is expelled. And on top of that she is calling him a name that does not legally or factually apply to Paul, and the university’sdecisions support the name is incorrect.
Therefore, back to my point, the university could tell her stop targeting another student and stop using an appropriate name fit him. That is not shutting down freedom of expression, that is protecting another student from verbal abuse of another. Her project can continue otherwise. That is quite different than your assertion I want schools to shut down free speech.
This is one of those scenarios where turn-around will become fair play, and the females on the receiving end would only have Emma to thank.
CF - It’s been many years for me and you may already know this. But “quasi-Poisson” isn’t a a statistical distribution. It’s sort of a trick. Suppose you look at your data and the counts have a much larger standard deviation than a Poisson model would imply (the variance of a Poisson model isn’t a free parameter, unlike even a Gaussian). This might be due to outliers (hopefully cleaned though) or even just because the Poisson model isn’t that good. So you futz with the likelihood function to “correct” for this. This is “quasi-Poisson” and it’s a standard option in many statistical software packages. Generally, the coefficient estimates don’t change but their standard errors are bigger, so the “t-stats” go down as you’d expect since the data is all over the place. Using just a Poisson model would be very bad because of the dispersion; the quasi-Poisson is really the first thing you should do to present the results in a non-misleading way.
A better way might be just to pick a better probability distribution (like negative binomial) but at some point you’re just torturing the data by trying to estimate more than you really can. For this student’s purposes it isn’t necessary.
Here is an introduction I found just from the first page of Google web.as.uky.edu/statistics/users/pbreheny/760/S11/notes/4-21.pdf
I just hope the moderator doesn’t ban me for talking about quasi-Poisson estimators!
The quasi-Poisson isn’t a distribution, because the probabilities of all the events don’t add up to 1 (nor, I think, do they add up to any finite number). So what’s the model here? What’s the point of modeling with a quasi-Poisson? What does it tell us?
Is it just that each event we’re considering has a likelihood, and even though the likelihoods aren’t probabilities (because they don’t add up to 1) we can nevertheless look at ratios of likelihoods?
But is a nightly hunt truly the norm? Seems like most kids are not booking up in numbers that big. Relationships seem to develop, but perhaps as kids get older. Certainly the hookup scene is much more prevalent now than in the 70s or 80s.
One interesting comment on the article was from a sorority women who stated that some sisters like the no men and no drinking rules as that makes the house a safe space for them. They can choose not to go to frat parties, but that choice would be gone if parties were held in their houses.
Also interesting that some women felt like they would not say something if they saw a guy hitting on a drunk sister, for fear of being labeled in a negative way. That is really unfortunate and really a place where education is critical for both - men for taking advantage of a close to incapacitated women and interfering with a friend who wants to help and the women for not being willing to risk social capital to stand up for their friends. I would think a similar dynamic is at play if lower status guys try to intervene - that may be the last frat party they ever get into (if not a member). Since the frats limit the non-member guys who get in to the parties, they certainly exclude guys who might be more likely to or have a history of intervene. Will bystander education work in these situations.
It seems like the more we learn, the more intractable the problem becomes. I hope I am wrong about that.
But here’s the thing @awc, by telling someone they can’t say something, no matter what that “something” might be, is in fact shutting down speech. Now you or I, having sons and possibly being somewhat sympathetic to the plight of Paul, might say it is the “right” or “decent” thing to do, but the university has to deal in practicalities. They have to have solid ground on which to make such a demand. That is especially so in the case of Emma who has proven to be quite persistent and has shown no tendency to back down when she feels an injustice has been done to her. So what would the universities solid ground be based upon?
I took a quick look at Columbia’s Rule’s of Conduct and there are only 2 sections that deal with speech. To violate the rules the university would have to find that Emma
So I think their rules are pretty much what I said in my post #856 and they would only be on solid legal ground if her speech threatened or incited violence. Pretty much what we have in the public at large, although I think there is also obscenity restrictions as well.
Now perhaps Columbia could come up with some other justifiable reason to shut it down, but right now no one is complaining to them to do so.
CF- Here’s a completely hypothetical example that might help - if you just use a Poisson, you might conclude that being in a high-status sorority leads to a higher number of hookups because the coefficient is positive and significant at a 99.99% confidence level.
But the Poisson distribution is wrong for the data, so you have to correct for this somehow, and the quasi-Poisson is a simple way to do it. You then find that the coefficient is only significant at a 30% confidence level. So your initial conclusion is not very well supported. Or maybe the confidence level is only 95% instead of 99.99%, so you’re still going to say that you have support but it isn’t as strong as the naive model would indicate.
The effect almost always works in this direction.
The bottom line is that reporting confidence levels with a Poisson distribution is almost always stupid when dealing with real world data unless you’re talking about physics or chemistry or you have some good reason to know that Poisson definitely works well.
This problem is very acute for the Poisson distribution because there isn’t a separate parameter for the standard deviation. A normal distribution at least has a separate standard deviation parameter. But linear regression significance levels can still be suspect due to things like heteroskedasticity, etc.
At least up until comparatively recently most studies published in even reputable medical journals didn’t even worry about this at all in a regression context That’s one reason (among many many) why all these medical studies that say that something is good or bad for you and they have a study that shows this with 95% confidence actually end up being wrong more than 50% of the time (unlike the 5% that you’d expect). Unfortunately, most doctors really aren’t numbers people.
“I would think a similar dynamic is at play if lower status guys try to intervene - that may be the last frat party they ever get into (if not a member). Since the frats limit the non-member guys who get in to the parties, they certainly exclude guys who might be more likely to or have a history of intervene.”
Actually, my understanding is that frats limit the number of reasonably attractive guys who are not members because they want the girls for themselves. Why invite competition? But, yes, causing a thing would probably keep you from being invited back. I do think some of you are over-complicating a very simple and old as the hills dynamic here which is probably part of why college administrator panels are a nutty and highly flawed idea to deal with college sex issues.
But if the quasi-Poisson isn’t a distribution, how does making claims about confidence levels even make sense? Confidence in what? There’s no underlying model here.
The verbiage I think you are referring to in the article actually turned my stomach. It’s hard to imagine that any of our girls would have to think in such terms before helping out a friend or any woman in need. I think that really hit it home to me that it is a cultural problem. Those sorts of embedded norms are hard to fight.
You’re right that there’s no explicit underlying probabilistic model. It’s a trick where you insert another standard deviation-like parameter into the likelihood function. There is sound justification because you’re starting from an assumed form of the likelihood function (which may or may not be invertible into a probability model. To be honest, I’ve forgotten and to reason it out would take me a while).
This enables you to calculate confidence levels directly using relatively weak assumptions - it’s not just a bunch of smoke and mirrors. I don’t know if you’ve studied this sort of statistics before, but if you haven’t then it’s not something that would be at all obvious just from probability theory.
Added: @HarvestMoon1 - I think you have a valid point about bystander intervention, but it’s complicated as I’m sure you realize. Despite the fact that we’ve been talking about sexual assault we have to remember that most sex is consensual. Maybe her friend doesn’t think she’s that drunk. Maybe she knows that she really likes the guy and wants to hookup with him and is just getting drunk so she has plausible deniability that she’s “easy”. Maybe after 99% these drunken hookups both parties are happy, so how do you know if it’s the 1% or not.
Maybe the cultural change you allude to is the right answer. But my understanding is that hookup culture is actually something that college women prefer at least as much as college guys.