Matriculation Rate Estimates for Science Programs & Take-aways. "Correlation vs Causation"
Hello. I've gathered a little bit of data. Wanted to know matriculation rates for many of these programs. I'll also talk about what you can conclude.
Honestly felt like I wasted an hour or two scrolling though and I realized I never cared about matriculation %%. Idk why I was worrying about this a few months ago. Everyone said that prestige doesn't matter, and they're probably right.
But the information is interesting! And it is a fun statistics challenge. Take everything I say with a grain of salt, I'm just a high schooler.
Note: data was gathered during different times, days apart. Might've miscounted some stuff :). I arbitrarily took the percentage of students who attend Top 15 college according to US News, and excluded LACs, Olin, International Unis. You can find this data on your own by looking up Instagram decision pages. LMK IF THERE ARE ANY ERRORS
COSMOS, Garcia, Clark Scholars, RSI, math/humanities programs not included due to little/no info.
Simons SRP.
| Year | Percentage | Response Rate |
|---|---|---|
| 2024-2025 (Instagram) | 16/21 = 76% | 20-30%?? uncertain |
SSTP Iowa.
| Year | Percentage | Response Rate |
|---|---|---|
| 2023-2025? (Instagram) | 30/43 = 70% | 20-40%?? uncertain |
BU RISE
| Year | Percentage | Response Rate |
|---|---|---|
| 2025 (Instagram) | 14/18 = 78% | idk |
CMU SAMS
| Year | Percentage | Response Rate |
|---|---|---|
| 2024 (Instagram) | 14/23 = 60.9% | ~30% |
| 2025 (Instagram) | 11/17 = 65% | ~28% |
percents look lower for SAMS but not statistically significant (barely) as sample is too small. Also because CMU is excluded as it is #20.
MITES Semester
| Year | Percentage | Response Rate |
|---|---|---|
| 2024 (Instagram) | 33/49 = 67.3% | ~20%? somewhat certain |
| 2025 (Instagram) | 48/61 = 78.7% | ~25%? somewhat certain |
Summer Science Program (2019-2021 were from official website ssp.org)
| Year | Percentage | Response Rate |
|---|---|---|
| 2019 (Census) | 111/140 = 79% | 90-100% |
| 2020 (Census) | 107/146 = 73% | 90-100% |
| 2021 (Census) | 126/168 = 75% | 90-100% |
| 2023 (Instagram) | 53/72 = 73% | ~35% |
| 2024 (Instagram) | 119/146 = 82% | ~39% |
| 2025 (Instagram) | 55/82 = 67% | ~14% |
| pooled census | 344/454 = 76% | |
| pooled Instagram | 227/300 = 76% | |
| pooled all | 571/754 = 76% |
Conclusion:
I think the Instagram posts are representative at least for SSP. The percentages are similar to the census, and there wouldn't be much reason for 2023 cohort being suddenly worse or better than 2021. The 67% for 2025 isn't statistically significant. When you do these tests make sure to account for multiple testing (ex. testing 2 hypotheses you should change a = 0.05 -> 0.025). I lost my data for the p-values :(.
All of the programs y'all put on the sub are great. Percentages are basically the same. I'd just go with whatever you like. The local internships at my area have like 10-15% matriculation rate lmao. Basically, Simons == SSP == MITES == SAMS == RISE == SSTP == probably every other program on this sub if you only care about matriculation (you shouldn't). I don't think a valid summer program ranking purely off of matriculation exists, and lots of kids here are just guessing. Same thing can be said for college itself maybe.
I'll answer some questions that the next year's summer program applicants may have and that I had a few months ago.
"Correlation isn't Causation? Does this data matter then?"
I still think this data is interesting. The statement correlation =/ causation doesn't automatically discard it.
I hear "correlation =/ causation" all the time online and in pop statistics. Though, I don't like when people just blindly copy and paste it.
In the context of summer programs and attending top colleges, "correlation doesn't mean causation" is a meaningless statement to the student inquiring about prestige and matriculation rates. Some of the people here are really analytical and information-seeking, which could be a good or bad thing. The main thing these students are looking for are the predictions for getting into top colleges (ex. What is the chance I get accepted to ASU?). They don't care about causation. Causation is irrelevant for prediction accuracy.
For example, a business may use the number of clicks to predict the number of purchases. A confounding variable could be buying intent. A viewer with high intent can click and buy more. The model still works and is useful, even if clicks and purchases have no causal relationship and are just strongly correlated.
Another example. Suppose your girlfriend really really wants to attend ASU. There's a program called RSI, and apparently it's an ASU feeder. She hears 70% of alumni from RSI get accepted to ASU, while ASU only has a 10% acceptance rate. If she gets into RSI, should she believe she has 10% to get in ASU, or 70%, with no other information? She should believe that she has a 70% chance of getting into ASU**, as she is given the information that she is part of RSI. The probability is P(ASU | RSI) = 70%. The only exception is if she thinks she is extremely different from the average RSI student for some reason.
So for those who got into top summer programs: great job. You've got great chances when you apply to college. Attending those summer programs may or may not cause acceptance to T20s. If not, they are still heavily correlated, and there is likely some quality of the students that attend those summer programs that causes acceptance. You can assume that you are likely to have that quality too that you did not know, because you got in.
** even this part is nuanced. One may ask why she should treat herself as a random sample taken from RSI alumni rather than a random sample taken from all applicants to ASU, and this gets into the unsolved issue of reference classes. A common resolution is to choose the most specific reference class, which is reasonable most of the time.
mb for the digression
"What do I do if I get rejected from these great programs?"
You can do volunteering or local internships; try not to do nothing over your summer. Look back at your applications and critique the essays that you wrote. Did your application have any red flags? Then work on common app/questbridge. Personal projects are cool too.
Also, your belief on the probability of getting into good colleges shouldn't change much from a summer program rejection. Only a fraction of students from these colleges even do summer programs. You also don't know the acceptance rate to top colleges of those who get rejected from summer programs, so there's not much information about your chances anyways. Don't stress about it.
"What do I do to get accepted to these programs?"
I would work on essay writing. Plus, it helps when you actually apply to college. Most of you guys have better ECs and awards than you think. But I don't think you should prepare for any of these programs.