College Acceptance/Matriculation Stats: NCS/STA, Holton/Landon

[Anonymous]

That is correct pp. Some people think that if they went to XYZ University; then they're kid will have no trouble getting in to it. That is not correct. Entrance/admission standards are much tougher now!

[Anonymous]

Yes, no longer automatic or an entitlement for some.

[Anonymous]

That's not what I meant. Even if the parents went to say UMD back in the day (which required no form of preference back then) and it was once easy to get into -- it isn't any longer. Why is everyone so angry on these posts always assuming everyone else gets some kind of deal or preference.

[Anonymous]

It's actually much harder to get in from a public school though it may be easier to stand out. For example, in my public school with a class of 550, only one student went to Harvard. He was straight A's, class president, and he invented something. Yet, there are prestigous private schools that are much smaller that send 40 plus kids to Ivies. I think that's why people are so crazy about certain schools.

While these top colleges are interested in diversity from different high schools, there are certain schools that seem to send an abnormally high percentage. And let's face it, there's a lot of public high schools out there. Harvard cannot even accept the top student from each school.

[Anonymous]

If nothing else, these threads demonstrate the desperate need for some of you people to take a statistics course.

In order to compare college acceptance probability from one school versus another, you need to control for both differences in the sample as well as differences in acceptance procedures of the colleges. If Harvard gives legacies a slight edge, that edge needs to be controled for. There is nothing wrong with things like legacy, money, even race entering into a college's acceptance decision. The question is, controlling for those factors, do certain schools in and of themselves increase the probability of acceptance to say a sample of highly competitive colleges.

Or are you asking does a school prepare a child better for Harvard, et al? In which case you need to control for the child characteristics at entry.

Or do you just want to compare matriculation statistics? In which case you are looking at numbers which have too much imbedded unobserved information to conclude anything.

[Anonymous]

Another history lesson for the ignorant. No one is angry here. There are the stupid, though, who require continual remedial education when they reveal the contents of their empty barrel. They, too, tend to make the most noise.

[Anonymous]

Anonymous wrote:In order to compare college acceptance probability from one school versus another, you need to control for both differences in the sample as well as differences in acceptance procedures of the colleges. If Harvard gives legacies a slight edge, that edge needs to be controled for. There is nothing wrong with things like legacy, money, even race entering into a college's acceptance decision. The question is, controlling for those factors, do certain schools in and of themselves increase the probability of acceptance to say a sample of highly competitive colleges.

And after you look at the graduating class, you need to ask where your kid fits in. If Harvard gives an edge to legacies, and half of your kid's graduating class at Sidwell are legacies (because Sidwell asks where the parents went to college on the application), but your kid is not a legacy, then going to Sidwell might work against you versus the local public. Not that there aren't plenty of legacy kids at a place like Whitman, but in some other local public high schools your kid might have a better chance.

Also, the person on the previous page had it right when she said that being at the top of your class at a top DC private could help you get into Harvard. But if your kid is in the middle academically at your top DC private, then Harvard isn't going to reach down to get him or her. The same kid might be better off being at the top of their public school (again with caveats -- it's probably tougher to be at the top in TJ or Blair than it is in even a top DC private).

So it's impossible to generalize about a kid's chances in public versus private high schools.

[Anonymous]

Anonymous wrote:If nothing else, these threads demonstrate the desperate need for some of you people to take a statistics course.

In order to compare college acceptance probability from one school versus another, you need to control for both differences in the sample as well as differences in acceptance procedures of the colleges. If Harvard gives legacies a slight edge, that edge needs to be controled for. There is nothing wrong with things like legacy, money, even race entering into a college's acceptance decision. The question is, controlling for those factors, do certain schools in and of themselves increase the probability of acceptance to say a sample of highly competitive colleges.

Or are you asking does a school prepare a child better for Harvard, et al? In which case you need to control for the child characteristics at entry.

Or do you just want to compare matriculation statistics? In which case you are looking at numbers which have too much imbedded unobserved information to conclude anything.

I think your point is that a school like Sidwell screens applicants for legacy status, high test scores in kindergarten, 6th or 9th grade, parental power and wealth (let's not discount that) and even recruitable athletic talent (I've seen it happen in two families). So that it's really no surprise that a high percentage of their graduates go Ivy. The same kids probably could have gone to an Ivy from the local public, given their test scores, legacy status and athletic talents. In fact, if I were an admissions director at Sidwell, one of my many goals (besides a balanced class, blah blah) would be to have kids who could boost the school's college exmissions numbers. But if your kid doesn't fit into one of these categories and got into Sidwell anyway, then they will be competing for those Ivy slots against all the other kids in their class who do fit one of the categories. Correct me if I've misinterpreted you.

[Anonymous]

I am saying that while descriptive statistics are nice, they won't answer the question about causality.

To do that, you need to construct a model of the probability of acceptance to Harvard (or wherever) controlling for the other factors that influence acceptance (such as legacy status, money of parents, individual characteristics of the student) and see what the effect of St. Albans (or whatever) has on that probability. That's if you want to answer the question of whether their is a 'school effect' -- that simply by virtue of having a school attached to your child's name, the probability of acceptance is influenced in one way or another.

Alternatively, you are talking about another issue altogether which is whether a certain school prepares a child better, thus increasing the probability of acceptance to Harvard. That's a bit more complicated, involving a 2-stage process whereby in the first stage you measure the qualifications most likely to determine acceptance to Harvard, and in the second stage the ability for a certain school to qualify a student along that vector of important characteristics, controlling for characteristics at entry.

[Anonymous]

PP here. I meant there, not their, above.

[Anonymous]

Anonymous wrote:

To do that, you need to construct a model of the probability of acceptance to Harvard (or wherever) controlling for the other factors that influence acceptance (such as legacy status, money of parents, individual characteristics of the student) and see what the effect of St. Albans (or whatever) has on that probability. That's if you want to answer the question of whether their is a 'school effect' -- that simply by virtue of having a school attached to your child's name, the probability of acceptance is influenced in one way or another.

While I'd love (really!) to see you do a probit model on this, I fear you'd never get reliable data to run it on. So then we fall back on observation -- who in the class at St. Albans goes to Ivy, and who doesn't?

[Anonymous]

Anonymous wrote:I am saying that while descriptive statistics are nice, they won't answer the question about causality. To do that, you need to construct a model of the probability of acceptance to Harvard (or wherever) controlling for the other factors that influence acceptance (such as legacy status, money of parents, individual characteristics of the student) and see what the effect of St. Albans (or whatever) has on that probability. That's if you want to answer the question of whether their is a 'school effect' -- that simply by virtue of having a school attached to your child's name, the probability of acceptance is influenced in one way or another.

I wonder if the analysis can't be a little simpler (although perhaps less precise). I don't think the goal is to create a probability analysis for any given individual student, which may involve dozens of relevant variables particular to that student. Instead, the goal is to figure out whether or not a particular school (or class of schools) increases the odds of admission. If you look at a large enough set of numbers, it seems that much of the "noise" brought on by individual student characteristics will drop out.

For example, while I agree that private school families are likely much wealthier on average than most public school families (because you're looking at families that can afford the tuition), I suspect there are relatively few with enough immense wealth to secure Harvard admission where a kid cannot qualify otherwise. And if you aggregate several years of data, the odds of having several of those immensely wealthy families skew the results drops even further. Athlete status also seems like it might be a relatively small percentage of each class, and I doubt most of these private schools are any more likely to field top athletes than public school students competing for slots. For legacy status, I suspect there's a greater chance of that threatening to skew data between private and public schools (if you assume private school applicants are more likely to be legacies than public school applicants). However, legacies account for only something like 15% of each class, so that's not likely to skew the statistics too much.

Maybe I missed the goal of your analysis, but it just seems like it doesn't have to be an impossible task.

[Anonymous]

Descriptive statistics are generally sufficient to point to a relationship, but in this case I think there actually is a lot of noise. Unlike something like a physical relationship or even a predictable behavioral relationship, you're talking about a decision-making process where there may not even be consistency from year to year. And certainly that process will be different from school to school.

I'm only saying not to read so much into the statistics, because there is much more to it than that.

[Anonymous]

Agree pp. Some parents think if they follow some statistical model; then their kids will get into Harvard. It doesn't work like that. We had 1 desperate mom at my daughter's elite hs. The mom did an excel spreadsheet indicating where kids from the prior 10 graduating classes had gone to college. Then she studied what activities they did. She made her daughter do all of those crazy activities. The girl did ultimately get into 1 Ivy, but not the one she was shooting for with her crazy mother. The craziest thing about the whole story is the crazy mom told people (bragged actually) about her excel spreadsheet. It did make a lot of parents either laugh or stress. Crazy stuff.

[Anonymous]

Anonymous wrote:Descriptive statistics are generally sufficient to point to a relationship, but in this case I think there actually is a lot of noise. Unlike something like a physical relationship or even a predictable behavioral relationship, you're talking about a decision-making process where there may not even be consistency from year to year. And certainly that process will be different from school to school. I'm only saying not to read so much into the statistics, because there is much more to it than that.

Good thoughts. But let's make sure we're talking about the same thing so I don't get too confused. In my post from 18:24, I was just focusing on the fairly broad question of whether a top private school gives any admissions advantage to admissions to top colleges over public schools (or alternatively whether students from public schools have an advantage, or alternatively whether there is no real difference between the two). I'm not focused on gauging any individual student's chances, and I think we'd quickly agree there are just too many particularized factors affecting each student to make any simple statistical assessment with the limited data we have available. I guess I'm focused on the general "relationship" question that descriptive statistics might answer, not on predicting any individual behavioral relationship.

Are you thinking about the same question? Or something more complex?