Elite college admissions: What matters and by how much

Parents Forum
1

A study linked in another thread on ( [Elite grads make more?](http://talk.collegeconfidential.com/discussion/comment/19987106/#Comment_19987106) ) gives me a hint: CC has the data and the talent to get to the bottom of what matters and by how much for a student to gain admissions to elite colleges. If rules allow, CC community can collect data publicly available on this site, analyze it, and report the results here. No one profits financially from this project, but everyone benefits from its findings. If at the end it turns out that character and virtue (wishfully) are likely to play the most significant role and every kid strives to reach the highest, then we all are better off.

Who are the elite colleges? Any college a CC member deems it elite. At the end we can analyze the top most popular ones in terms of data collected, and the elites can be top 100, 50, 20, 10, 5, or even 1 if there are enough data.

How to analyze the data? Whoever proposes a method, show us and let us refine it. Better yet, use your method and present your findings and let us be the judge.

What data to collect? Nothing better than a student who applied to multiple colleges. I imagine for a student, his/her data for 1 college is a row with many columns that include college name, outcome (accepted (EA/ED/RD)/waitlisted/rejected), test scores (SAT, ACT, SAT II, sub scores), GPA (weighted, unweighted), class rank (known or presumed), AP/IB (subject and score), EC/job/summer activities (what, roles, length, awards), counselor evaluation (known or presumed, private or school), essay and supplement short answers self evaluation, interview self evaluation, race, gender, state, school (private or public), family income, parents highest degree, self reflection of why the outcome, and lastly the CC data collector’s impression of the student’s clarity and conciseness (1-10 scale). A few missing values are ok. What else can we get?

Anyone knows how to put the data in a file? Is it possible here to do what Wikipedia does where anyone can access and add/edit? Maybe CC has already collected the data and it would be nice to share with the community with a link.

Data and algorithm can be used to find a match (see [Amy Webb did for love](http://www.npr.org/2014/04/25/301822006/can-you-use-algorithms-to-find-love) ).

2

Feel free to go through college admissions results threads on each college’s forum section.

3

Data of this nature will always be flawed because you are not taking into account key components that factor into holistic admissions at a high level: essays/supplements, teacher recs, counselor recs. We already know the data: get yourself above a certain threshold, and you are considered. Beyond that, your essays, recommendations, in some cases whether you can pay for the school, hooks if you have them… heck, what the admissions officer ate for lunch and whether it put them in a bad mood, will determine whether you’re accepted. It’s also always relative, each year, to the rest of the applicants and where you fall along the spectrum of putting together a diverse and well-rounded class. You can’t quantify it.

4

The going CC dogma is that no one can ever know what gets a kid into an elite college. I suggest not stirring the hornet’s nest.

5

Can some CC members write a program to extract all the info and put in a file? Manually it needs manpower. If I were to do it myself, it could take me years to learn statistics and computer programming, collect the data and do the analysis.

We can put it in the basket of “unknown” and quantify it by what’s known.

Box said “All models are wrong, but some are useful.” Hopefully we can see how much the unknown counts, 20, 40, 60, or 80%?

I think CC parents accept what data says, to the extent what the data can say.

6

Useless. Going out on a limb, there would be lots of missing data from applicants who did not get into their dream schools. People have been known to lie on surveys. Shocking, I know.

7

To amplify what periwinkle just said above, even if you went through EVERY results post out here, you wouldn’t even scratch the surface of all the admits/denials at these schools.

A handful of schools report specific breakdowns of scores, GPAs, etc (I’ve seen this from Brown and Georgetown specifically). It’s pretty easy to see what is important to the school. But even if you’re perfect with your stats, you only have about 30% chance of admission. As others have said, with holistic admissions it’s impossible to see exactly why someone got in. But we all agree that you at least ‘have a chance’ with stellar stats and something hook-worthy.

8

So you want somebody to volunteer to do all this for you so you can use it for the benefit of your own kid. I can’t see a lot of people lining up to do a gargantuan task for free. Bottom line: your child needs good grades and good test scores to get into elite colleges. Beyond that, there are a million variables. You can use one of those paid websites to more accurately predict your kid’s chances. For the same money though, your child can just submit an application to the elite college of his/her choice. Now start saving up for private school.

9

Awhile back I followed a similar approach with the Stanford decisions thread on this site. I found that the vast majority of decisions could be predicted correctly by looking at only out of classroom activities, such as ECs and awards. Academic stats actually had a very slight negative correlation with admission. I expect this result related to the CC decision thread posters being a very unique subgroup who tends to have outstanding academic stats, but has more variable ECs/awards. As I recall, the majority of thread posters were accepted, yet the overall acceptance rate across all 40k+ applicants was ~5%, which also suggest an extreme bias in the subgroup of CC thread posters. In short, I think the CC thread posters are such a small and biased subgroup that you won’t be able to gleam much about admission for typical applicants.

Parchment used to has a database like you describe. They used to allow sorting the database by a wide variety of criteria. For example, you might choose to compare acceptance rate for different genders among applicants with similar academic stats, similar number of AP classes, etc. I found it useful. However, they no longer appear to support this and instead just have generic scattergrams like you’d find on Naviance. Furthermore Parchment has some serious issues with the quality of information due to the self reporting nature. For example, the current Parchment rankings of which colleges students who are accepted multiple colleges choose rank Harvard 22nd, between Pitzer and Texas A&M, suggesting students who are accepted to Pitzer, Harvard, and Texas A&M have about an equal chance of choosing to attend any one of them. This occurs because some Parchment members claim to get in to Harvard with GPAs as low as 1.9, then choose to attend far less selective colleges instead.

10

Here is something funny: It has already been done, and published!

https://www.amazon.com/s/ref=nb_sb_noss_2?url=search-alias%3Daps&field-keywords=Behind+The+Ivy+Curtain

from the book’s description:

*Want to know what separates successful Ivy League applicants from the rest? This book uses real data from nearly 5,000 students to discuss what works and what doesn’t for Ivy League admissions, and presents a plan that any student can use to maximize their chances of getting accepted to a top school.

**I analyzed SAT/ACT/GPA/test scores as well as qualitative things like extracurriculars and recommendations/essays from the College Confidential forums to build up a database of profiles and admission outcomes. *The book is intended for high school students and parents who want to learn more about what happens “behind the scenes” in the admissions process and what things make an applicant competitive and what is just a waste of time.

I have the book, I will tell you what I think of it if you like.

11

I read the book. No one outside the admission office has a perfect database; neither does the author. The author acknowledges this. Nevertheless, I found his analysis to be quite careful and I agree with most of his interpretations of the results. Subsequently, I ordered another copy of the book and gave it as a gift to a friend of mine who has a kid aiming for Ivy admission.

12

Useless quest. I’ll bite. Perfect SAT scores do not mean getting into the most elite private schools. Gifted kids often may have less than perfect grades. They also may not want to apply (regardless of parental suggestions) to some supposedly tippy top schools. Flawed data collection already. Who knows where son could have gone at age 16? Top tier rankings in many fields for state flagship made it any easy choice- and hard to justify applying to many privates.

Too many variables. I guess you could say that student interest plays a role (see above). Then look ten years into the future and see if going to school A, B, C, D… really makes a difference.

13

Also @prof2dad, do you two have his data? I ordered and read his book and requested the data as he suggested but not yet received it. Either he’s too busy or changes his mind. My definition of analysis may differ from yours. His looks more like a qualitative analysis to me but the results agree with what posters here have been saying. So I think the posting HS students were truthful. We can go deeper analysis-wise IMO.

14

I missed your analysis, due to my being a recent and infrequent CC visitor? I think your finding is very important that the most powerful predictor is out of school activities which may be related to something deeper and thus ungameable. I’d like to peek into the unknown where maybe the core of holistic review lies.

15

I did not request the data. But the summary statistics reported in the book are quite consistent with my casual observations on this website.

16

Oh man! That book won’t make many friends in the CC circle IMO.

17

Despite the offer within the book, my requests for the data went unanswered.

18

I’m collecting data and it’s toooooo slow, discouraging, not fun. But I have a feeling that, as Data10 puts it, the key lies outside of school work. We shall see what they are and how much each weighs. In the meantime, parents of students capable of SAT >2200 should encourage them to exercise more and rest well.

19

It’s called Naviance.

It doesn’t matter that a kid who grew up on the South Side of Chicago got into Yale with a 720 verbal and 740 math score. MY kid- from a northeast HS which sends a high percentage of kids to elite colleges, would NOT have gotten in to Yale with those scores. It doesn’t matter that a Val from a rural HS in Wyoming got into U Chicago with “no EC’s” except for working the family farm and helping to take care of an elderly grandmother. No kid from a private school in the LA area will get into Chicago without some substantial and meaningful accomplishment (yes, farming counts but I’m not aware of working farms in LA right now). It doesn’t matter that a kid who has spent four years bouncing around from homeless shelters to public housing in NY got into Harvard with a 92 average, 700 SAT scores, and only one rec (from a social worker- not a GC since the kid has been in four different HS’s and none of the GC’s know the kid). YOUR kid isn’t getting into Harvard with a 92 average and no rec’s from a GC and two teachers.

This perseverating focus on data is strange. Exactly what do you hope to learn- other than a 9% admission rate means tough odds?

20

Any data analysis is about getting a big picture. By construction, getting a big picture requires not taking extreme observations too serious, sometimes we even truncate those 5% most extreme observation out of the dataset.

I do like the idea of having kid from rural North Dakota for diversity reason. But this observation is simply not a part of the big picture. Via data analysis, we may learn something (surely not everything) for say 80% of elite applicant population, and surely will not do so for some others say 5%.