Chance your own danged self! (Here's how)

Chance Me / Match Me!
21

@Corinthian Post #18

I am always somewhat surprised how much identifying info some kids put into these posts. At least of couple of times, I have been able to find out their names in a few seconds on Google.

On the larger point though, I doubt that many (or any) schools have the resources to scour these forums for this stuff.

With respect to issues like “Tufts syndrome” or related problems with safety schools, I think there are lots of ways to show or express enough interest (visits, emails, well-written “Why ?” essays, etc.)

22

Thanks so much the clear game plan! I especially like #7.

You have the grades, test scores, and rigor of the typical kid at XZY college? Great, you have the Admit Rate chance of getting in, which for most Chance Me posters means go find some safeties, and no 10 schools with 10% admit rates does not equal safe. Now, if you can just come up with a magic formula to tell me how much money colleges are going to throw at my baby girl. :wink:

As for your post title, I’m just waking up here in Hawaii and hadn’t had my first cup of coffee. I thought it said “deranged” not “danged.” Haha, a little harsh @NickFlynn!

23

@palm715

“Now if you can just come up with a magic formula to tell me how much money colleges are going to throw at my baby girl”

Actually, that is what I really am working on, but it is nowhere near as developed or reliable as this stuff yet.

24

If your college ranking is based on getting the enrolled/accepted student ratio as low as is can get, I think it is naive to think that there are not admissions officers trolling the forums.

Paranoid? Yes.
Improbable? No.

25

Above, I promised to discuss some methods to compare your GPA to the target schools data, including how to estimate the school stats when there is limited info provided. I started that discussion, and the link is:

http://talk.collegeconfidential.com/college-admissions/1795818-convert-your-gpa-to-sat-score.html

Note, the method I am using there is somewhat different from what I showed in the first couple of posts on this thread.

Apologies for the shameless self-promotion, but in my defense, I did receive a lot of positive feedback from this post, so I assume at least some people are finding it useful.

26

I never saw this and wow this is like the most helpful post I have ever seen on CC. Well done!

27

@palm715 Don’t most colleges actually give you calculator-type thingys where you can put in requested information from tax stuff and then they give you an average based off of it? (Example: https://saservices.berkeley.edu/calculator/ the CAL-culator for Berkeley).
It’s different for each college so you’d have to spend quite a bit of time gathering all the info and plugging it in, but isn’t that basically a formula where colleges tell you how much money they will give you?

28

I’m sorry but to me the whole idea of “Chancing” someone beyond saying "You appear to be a competitive (or non-competitive) candidate is typically a waste of time since there are SO many factors that go into making an admissions decision. As we know, there are many non-statistical factors that enter into the holistic process.

This whole method described above which is seemingly quantitatively based on numerical data is actually fraught with subjective judgements such as the the arbitrary weights assigned to course schedule to determine the “rigor” or the “Multiples” that are applied to the percentages (calculated by averaging the GPA and test scores - where did this equal averaging even come from let alone the “Multiples” to be used thereafter?). This all reminds me of how they run the 10 yd chains out during a football game and place one end down on a fairly arbitrary spot so that they can achieve “precision” on the other end.

We are invited to “tinker” with the “formula” (this is really a formula?) but how are we supposed to know how to make any judgements on how the formula should be “tinkered” with since there is an absolute lack of any meaningful data to help us run any analytics to do so.

Even using the 25%- 75% ranges as starting points for the GPA and test scores is a complete waste of time because we know nothing of the composition of admitted students in these ranges. For instance, suppose the bottom quartile are 100% green-eyed brunettes (not going to even say URM’s, athletic recruits, first generation + economically disadvantaged, children of development cases, etc as is widely speculated). How would this “model” even begin to account for that? Could it tell you that you are wasting your time applying as an “unhooked” blue-eyed blonde?

A key factor in your model is the admit rate, but the number of foreign applicants (particularly from China) has skyrocketed in the past five years. At most top schools, the number of foreign attendees is capped around 10% so the portion of the declining admittance rates coming from increasing international applications actually has no bearing on US applicants. How does your model account for this?

Finally, you tell people to Chance their own “danged” self using this model. The fact is, they can easily look up where their GPA’s and SAT scores stand in relation to the admitted class and get a rough idea of how they stand numerically. It’s all the subjective things which you list that they are wondering about and are looking for outside interpretation/validation.

Your model ignores these things. Intel STS winners, world class musicians, accomplished authors are all given the same percentage “chance” as a couch potato with the same grades and test scores.

You say you factor things like URM status into your spreadsheets at home? How do you even begin to do this since again the data are completely unavailable to create any kind of useful factor. You can’t even try using years of CC data because there is no way to even verify if what people are posting are correct information. Many kids post wrong (usually inflated) statistics to either hide their identity or massage their egos.

Finally, there is no way to either backtest your model or even see how accurate it is going forward because again there is no way to obtain any necessary data to do so.

Therefore, I find the whole exercise interesting but of little real value, I’m sorry to say. Kids are better off hearing from random internet strangers because at least they understand the chances being thrown around are completely subjective and so won’t be lulled into thinking there is any real statistical basis for percentages being generated by a “model”.

29

@Falcon1

Everyone is entitled to their own opinion - it does seem like there were other people who thought this was useful.

My own opinion about your response is that it is at least partly motivated by a disagreement we had on another thread, but really that is neither here nor there in the bigger picture. I am not interested in having an “internet board feud” with you.

I would respond in detail to your criticisms, but I have reason to believe that you will take that as a sign that I want to debate every little point with you endlessly. If you think what I presented here is useless, then I respectfully suggest you move on and ignore it. Seriously.

The whole idea of developing a model is to make explicit your assumptions. When you develop of model of a complex system that includes subjective factors or factors that can’t really be measure, you can still add considerable value by simplify quantifying the parts of the system that can be quantified. Moreover, there is ample anecdotal evidence that the quantitative parts of the model in fact track pretty closely with the initial classification of applications by admissions officers (sorting applications into bins or pools), so the creation of such a model passes the “common sense” test.

And, as for the other stuff (the unmeasurable and the subjective), I am being totally upfront about what those factors are, what those limitations are, and did nothing more than suggest ways that those factors can be estimated.

As for the claim that this framework is useless (“kids are better off hearing from random internet strangers”) and unverifiable:

  1. Because I am making numerical predictions, anyone who cares to could potentially followup on my predictions and see if they have any validity statistically.

  2. In fact, I have used variants of this system to chance kids that I have worked with during the application process, and the results have in fact tracked pretty closely with predictions.

  3. I have also used some of the results threads here on this site to cross check against my own estimates. In this trials, the system has proved to be reasonably reliable. Obviously, small sample sizes + unknown data quality. Also, I have learned lessons from doing this (the biggest and most noteworthy being the need to incorporate GPA + course rigor information and weight it evenly with test scores) and those lessons have led to changes in the model to improve accuracy.

  4. I have discussed the system in extended conversations with several professionals in the field (adcoms and counselors) and the response has been highly positive.

You obviously disagree, and you are certainly entitled to that opinion. Everyone has the right to decide for themselves. I put this out here because I thought that some people would find it valuable, and I have gotten some positive feedback, so that wasn’t pure fantasy on my part.

Recent experience with you tells me that nothing I say here will discourage you from trying to turn this into an argument, but I am not interested in a debate with you.

30

I'm would not be insulted if you spelled out the exact formulas. I am interested in knowing how they are arrived at. For the factor weights. It is impossible to run multiple regressions because there is zero data to do this on.

I have never seen or heard of this and I so I am eager to see the evidence. Please list some of the sources for it.

I already cited how the admit rate is being skewed by the sharp rise in international applicants. Also, huge outreach efforts to socio-economic disadvantaged kids particularly in the Midwest has been generating increased apps for many of the very top schools. Most of these kids are not viable candidates though, so this is also lowering admit rates but may not have any real effect on competitive applicants .

31

@Falcon1
Against my better judgement, and because I think it is an important point worth emphasizing, I am going to respond.

  1. Use BOTH GPA and Test scores…

Well, the first piece of evidence would be the CDSs of the elite colleges - almost without exception, they rank GPA, course rigor, and class rank at the same level of importance as standardized test scores (in fact, a few rank test scores below those factors.) I am willing to take them at their word.

The second piece of evidence would be that the GPA numbers of the admitted students (for the colleges that disclose them) are generally even more tightly bunched than the test scores. That is indirect evidence at least that the schools care a lot about HS GPA.

Lastly, there are many studies that show HS GPA correlates better than standardized test scores with college outcomes such as graduation, retention, and first year GPA. The general consensus is that this is the case (some of the studies are even by the College Board and the ACT folks, so that lends additional validity) and so it would be kind of strange if admissions committees would not be aware of this fact.

As for the point about admit rates, then I have to ask (actually, rhetorically, so you needn’t respond), is it your position that admit rates are NOT an important consideration in assessing an admit chance?

Are admit rates subject to some distortions? Are certain schools more prone to this than others? How do you measure that? Interesting questions all, I suppose, but probably nearly impossible to measure with any accuracy and it is entirely feasible that when comparing similarly ranked schools, any effect may in fact be small enough to be safely ignored.

32

WOW, NONE OF THIS even remotely suggests that Adcoms EQUALLY WEIGHT the importance of grades and a standardized test score such as the SAT the way you do by assigning a 50% to each!

And how is this even possible? If you tell someone they have 13% chance of getting accepted to Harvard and they either get in or are denied, how do they measure the accuracy (or “statistical validity” as you put it) of your “predictions”.

Years ago, I co-headed the statistical arbitrage dept. of a major Wall Street investment bank. What you are trying to pass off as a viable model (with a complete lack of usable and clean data, subjective factor weights, and no accurate way to backtest or even real-time verify the signals) would not even last five seconds under scrutiny.

All right, never mind, I won’t bother arguing with you anymore. If you think what you are doing is helpful, all the more power to you. Buyer beware.

33

MODERATOR’S NOTE:
At this point, let’s agree to disagree. People are posting their opinions and other readers are free to take from it what they will. I will assume that there will be no further back and forth in this thread between 2 posters.

34

Am I missing something?

I agree with @NickFlynn that the “chance me” posts are somewhat silly (and if they disclose too much identifying information they can be unwise). But I also don’t think it’s accurate or helpful to use this kind of formula to “chance your own danged self.” I agree with @Falcon1 that it’s misleading to imply that such a formula would be a meaningful tool.

Why do we even need to see percentages? (Seriously, that’s not a rhetorical question.) I understand the human urge to have them, but it seems to me they are meaningless.

The percentage chance can’t possibly be accurate. Unless you are an insider and modeling with respect to your school only, any model is going to be extremely subjective and very limited by its inability to account for all sorts of factors. Even if you’re an insider, the model won’t account for essays that grab or turn off particular readers, for example.

At the end of the day, isn’t the point to figure out, first of all, which schools can be considered “safeties.” It’s most important to apply to at least one true safety school. After that, it’s nice and somewhat helpful to figure out which schools are possible “matches,” which are “reaches,” and which are “far reaches.” It’s nice to know, so that you don’t load up on all reaches and far reaches, but really, isn’t the point to select a range? It doesn’t even matter a whole lot if you get one or two (or even more) categorizations wrong (classify a school as a “match” when it’s more truly a “reach”), so long as you have a range. If you’re in love with a school, for gosh sake, go ahead and apply, regardless of chances. But don’t hang your hopes and dreams on any one or two schools.

I can see that one might change one’s behavior based on whether a school is deemed to be a far reach versus a match or maybe even a reach versus a match, but what does a particular percentage chance matter? Any finer-tuned analysis beyond safety, match, reach, far reach must be meaningless because there is no way it could be accurate with respect to any individual and irrelevant because no one should change any behavior based on the particular percentage that the model supplies. Am I wrong?

35

Well, the last thing I want to do is to provoke another argument, but some of this I agree with, but some of it I don’t.

The “method” that I presented here is basically taking the objective factors that we know about the previous incoming freshman class (test scores + GPA) and then seeing where an applicant falls along that spectrum. There’s absolutely no rocket science or secret sauce being used there, it’s simply basic arithmetic and there you are. You don’t have to take it any further than that. You know where you stand on those objective criteria versus the data that the colleges publish. Obviously, that is not the whole story, it is just part of the picture - there are all the subjective factors of your application - recommendations, essays, supplements, extra-curriculars, interview, etc.)

However, if you stop there, you are missing another big piece of the puzzle. Having test scores and grades right in the middle of the pack means something a lot different if you are talking about Harvard versus some mid-level public university. That’s why I think the next step - using the admit rate and your relative rank on the objective factors to generate a “chance” is important. That’s what tells you that Harvard is still a reach, and the public is maybe a match.

If you look at the simplified method I outlined in the original post, you can see that everyone gets sorted into only 5 quintiles, so it is not like there is any implied hyper-accuracy in the system that is wildly unrealistic.

Here is the main point about all this though that I think DOES make it useful for people as a decision tool. All the soft application factors that can’t be estimated (the list above) - they are all going to be evaluated roughly in the same way at similar schools. They are, after all, the same stuff mostly, and the adcomms probably will react in similar ways to them. So, all that stuff essentially becomes a constant - and that can make the relative differences between schools in terms of the objective factors more determinative.

If the method says you have a 10% chance of getting into Harvard and 25% chance of getting into Georgetown - that does tell you something valuable EVEN IF the absolute chances aren’t dead on right. Depending on what the rest of your applications look like, you might choose one or the other to fill out your slate of applications.

Lastly, when you get to situations like considering the impact of gender or race or legacy or early decision versus regular, etc - there is information out there about all of these things, and you can use that to make the system more accurate for people who fall into these categories. Is this information 100% reliable? No. Is it possibly handled slightly differently by different schools? Yes. Does that make it not useful? Absolutely not, in my opinion. You see this kind of question come up all the time on this board - “I’m a black male with these scores and grades - is my list of schools reasonable?”

I absolutely don’t expect that everyone is going to see it the same way I do, and that’s fine. I also don’t expect people to just accept that this is bestest chancing methodology that exists - it’s definitely not, and it is a very simplified version of what I actually use myself, but I do think it is a good framework. I posted it because I thought some people might find it useful, and also so when I chance someone in a thread, if they ask me where I came up with the numbers, I can point them here.