Hey everyone!
I’m Matt -- I do data science here at CollegeVine and have been around in the past talking about some of the methodology behind our chancing model (In addition to name dropping Carleton College when I get the chance)!
We’re upgrading everyone to a new version of chancing in the next week! This is the best version of CollegeVine's chancing model yet. We are expecting everybody’s chances to change a bit, so please don’t panic if you log in and see different chances on your school list :). Here are a few highlights of the newest chancing version:
1. We’ve added a new input to the model! We’re beginning to account for the effect of applying early decision on your chances, so if you plan to apply ED, expect to see your chances increase slightly.
2. We’re now using forecasted acceptance rates in the model. We’ve received feedback from students about how the model cannot be correct because most acceptance rates are trending down over time. The newest version of the model builds these trends directly into your chances. Instead of using last year’s acceptance rate as a baseline to predict your chances next year, we predict what next year’s acceptance rate will be based on past data and use that instead.
3. As is typically true with our model upgrades, the data underlying the model has been updated to incorporate the latest available information.
We’re really excited for these changes to go live to all of you! If you see anything unexpected or have any questions, I’d be happy to do my best to answer if you drop a line here. Or as usual, feel free to reach out to support@collegevine.com. You can also check out this blog post for more on the methodology behind the model!
Matt
hey @sophiebeatriz, you can see them if you refresh!
A lot of my chancing dropped significantly--what is the most probable cause? As in what specific part of the update would have most likely caused it?
(1/2) hey @i-marie, there are two components that were probably the causes: one is that we made some significant methodological improvements under the hood for how we determine how your academics (GPA, tests, course rigor, etc.) stack up against the academic profiles of accepted students at the colleges on your list. It was a big change, so that could definitely have a big effect.
Matt, I think its about time someone at CV updates all the baseline admission rates for the colleges in the CV database. When you curate a list of colleges under "MY LIST", you see very old admit rates some are from 3-4 cycles back and are no longer valid. Also, it gives poor context to the new version of chancing % as well if the old ones are meaningless. For example:
Carleton is 16% not 21%
Barnard is 8% not 14%
WashU is 10% not 16%
Tufts is 9% not 16%
Middlebury is 14.9% not 22%
Washington and Lee is 18.5% not 25%
NotreDame is 12.9% not 19%
Vassar is 19.5% not 25%
Colgate is 12.3% not 27%
NYU is 12.2% not 21%
Boston U is 14.1% not 20%
Boston College is 16.5% not 26%
Hamilton is 11.9% not 18%
Wellesley is 13% not 20%
Emory is 10.7% not 19%
Caltech is 2.5% not 7%
Yale is 4.5% not 7%
Harvard is 3.2% not 5%
Pomona is 6% not 9%
Swarthmore is 7% not 9%
Princeton is 4% not 6%
Columbia is 3.7% not 7%
Brown is 5% not 8%
Colby is 7.5% not 10%
Duke is 5% not 8%
Amherst is 7% not 12%
MIT is 4% not 7%
Dartmouth is 6.2% not 9%
Northwestern is 7% not 9%
NorthEastern 6.8% not 20%
Williams is 8.5% not 15%
JHU is 6.5% not 11%
UPenn is 6% not 9%
UCLA is 10.7% not 14%
Cornell is 6.9% not 11%
Vanderbilt is 6.1% not 12%
I do think its a good idea to use forecasted rates and the ED/EA feature will be a good "add".
This update seems likes it’s completely ruined the chancing statics. Any school ranked 1-20 is about a 1-9% chance of admission. Any school 21-25 is 44-65% chance of admission and any school ranked 25-100000 is a 70%+. I have no idea what you went about changing but most of my admissions stats dropped dramatically or increased dramatically which makes me really question the legitimacy of your service.
Just a quick question:
My chances for several of my more competitive schools dropped significantly:
Vanderbilt: 25% to 17%
Duke: 18% to 8%
ND and UVA experienced significant declines as well.
Does this have to do with the update?
Agreed with @CameronBameron
Hey @matt.kaye!
Any plans to add an application cycle dropdown to the chancing simulator so we can compare mid-senior-year GPA estimation chances to that of applying early? That'd be awesome.
I meant to comment this... my bad.
To keep this community safe and supportive:
How do I see how early decision will affect my chances? The normal chancing doesn't change when I select RD or ED. Is this feature just not available yet?