Post Interview Acceptance %'s

[RogueBanana]

Deleted.

---

Members don't see this ad.

 

[787593]

.

 

[aalamruad]

*poof*

 

[779163]

Here's the link to the color-coded spreadsheet: Link

 

[RogueBanana]

Deleted. Incorrect stat calculation.

 

[aalamruad]

Interesting,

If we take the average of all Post-II acceptance rates (from the google doc you linked me) we get 50.27 %

This may explain the "interview's three, a doctor you'll be" rule. 2 IIs gives a tentative 100% chance, the additional II controls for variability between schools. That of course assumes you are a decent interviewer.

2 IIs gives a 75% chance of acceptance with 50% post-interview acceptance rates

 

[aprilfools]

Interesting,

If we take the average of all Post-II acceptance rates (from the google doc you linked me) we get 50.27 %

This may explain the "interview's three, a doctor you'll be" rule. 2 IIs gives a tentative 100% chance, the additional II controls for variability between schools. That of course assumes you are a decent interviewer.

Not quite. The probability of being accepted after N interviews with an average acceptance rate of ~50% would be 1-((1/2)^N), not simply the additive sum of 50%+50%. For three interviews then, you'd have an acceptance probability of 87.5%.

 

[RogueBanana]

Not quite. The probability of being accepted after N interviews with an average acceptance rate of ~50% would be 1-((1/2)^N), not simply the additive sum of 50%+50%. For three interviews then, you'd have an acceptance probability of 87.5%.

Thanks, my stats are a little rusty lol

 

[Henchman21]

Much appreciated! I don't see Tulane though. Do they not publish the number they accept?

 

[aymar]

This is gold. Thank you for compiling it!!

 

[PB&Jam]

Ouch. As someone hoping for some love from Penn in March, this doesn't make me too optimistic :/

 

[RogueBanana]

Ouch. As someone hoping for some love from Penn in March, this doesn't make me too optimistic :/

I'm sure you'll get some love.

 

[ScreenName23]

Is that Penn State or UPenn? @RogueBanana

 

[RogueBanana]

This is gold. Thank you for compiling it!!

Happy to help, I always wondered why we didn't have a thread like this.

 

[RogueBanana]

Is that Penn State or UPenn? @RogueBanana

I believe it is penn state

 

[779163]

Not quite. The probability of being accepted after N interviews with an average acceptance rate of ~50% would be 1-((1/2)^N), not simply the additive sum of 50%+50%. For three interviews then, you'd have an acceptance probability of 87.5%.

These aren't independent probabilities, though; probably the same groups of people are getting accepted at a lot of these schools.

 

[aalamruad]

Is that Penn State or UPenn? @RogueBanana

I believe it is penn state

I don't think it's penn state, the OP on the reddit post said s/he didn't have any info for penn state

 

[RogueBanana]

These aren't independent probabilities, though; probably the same groups of people are getting accepted at a lot of these schools.

I interpreted it as the probability of one person being accepted over the course of 3 IIs

So if we interpret it as a coin flip (p= 0.5) and the events are mutually exclusive, you sum the probabilities.

1/2 + 1/2 + 1/2 = 3/2 =150%

Am I misdefining the events in terms of independence?

 

[aalamruad]

These aren't independent probabilities, though; probably the same groups of people are getting accepted at a lot of these schools.

True, but there's not really any way to factor that in that I can think of

I interpreted it as the probability of one person being accepted over the course of 3 IIs

So if we interpret it as a coin flip (p= 0.5) and the events are mutually exclusive, you sum the probabilities.

1/2 + 1/2 + 1/2 = 3/2 =150%

Am I misdefining the events in terms of independence?

You can't have >100% chance of something happening. I think your confusion comes from independent vs. mutually exclusive. They are independent events, but not mutually exclusive. They're independent because one event (acceptance/waitlist/rejection) doesn't affect the other admissions decisions, but they're not mutually exclusive because one acceptance doesn't mean you can't have another acceptance, and one rejection doesn't mean you can't have another rejection.

 

[779163]

True, but there's not really any way to factor that in that I can think of

You can't have >100% chance of something happening. I think your confusion comes from independent vs. mutually exclusive. They are independent events, but not mutually exclusive

There isn't a way to factor that in--but I do feel that if you are not a very good interviewer, you should not look at 87.5% chance of acceptance and take comfort in that. You need to go and work on your interview skills. Thinking about chances is useless

 

[RogueBanana]

True, but there's not really any way to factor that in that I can think of

You can't have >100% chance of something happening. I think your confusion comes from independent vs. mutually exclusive. They are independent events, but not mutually exclusive. They're independent because one event (acceptance/waitlist/rejection) doesn't affect another admissions decision, but they're not mutually exclusive because one acceptance doesn't mean you can't have another acceptance, and one rejection doesn't mean you can't have another rejection.

ohhhhhhh I get it, thanks!

There isn't a way to factor that in--but I do feel that if you are not a very good interviewer, you should not look at 87.5% chance of acceptance and take comfort in that. You need to go and work on your interview skills. Thinking about chances is useless

I'm not saying we should be complacent based on probabilities, just remarking that an old adage may have some statistical basis behind it. Though it doesn't take a statistician to understand that more interviews = greater chance of acceptance

 

[779163]

ohhhhhhh I get it, thanks!


I'm not saying we should be complacent based on probabilities, just remarking that an old adage may have some statistical basis behind it. Though it doesn't take a statistician to understand that more interviews = greater chance of acceptance

This poll might be unreliable because of sampling bias, but related to what you are saying: https://forums.studentdoctor.net/threads/interviews-three-a-doctor-youll-be.1215042/

I just think people should be cautioned against interpreting 3 interviews as having "good odds" given that many of these acceptances will be held by the same group of applicants. A better, albeit still flawed, way to think about it would be calculating the odds of matriculating at one of the schools that you are interviewing at.

 

[efle]

TIL that as long as you post it to Reddit before SDN you can put up stuff that is normally behind paywalls...

 

[PB&Jam]

I believe it is penn state

I don't think it's penn state, the OP on the reddit post said s/he didn't have any info for penn state

Yeah Penn=UPenn, I kinda felt silly on my interview there when I realized they refer to themselves as Penn and I kept adding a U

 

[Gurkhali]

TIL that as long as you post it to Reddit before SDN you can put up stuff that is normally behind paywalls...

It's statistics laundering. Reddit is just a front. The SDN mafia is still at large. More at 11.

 

[nejikon]

Do the accepted numbers include the waitlist?


Sent from my iPhone using SDN mobile

 

[aprilfools]

These aren't independent probabilities, though; probably the same groups of people are getting accepted at a lot of these schools.

Correct, but we're just talking about theoretical odds for arguments sake here

This poll might be unreliable because of sampling bias, but related to what you are saying: https://forums.studentdoctor.net/threads/interviews-three-a-doctor-youll-be.1215042/

I just think people should be cautioned against interpreting 3 interviews as having "good odds" given that many of these acceptances will be held by the same group of applicants. A better, albeit still flawed, way to think about it would be calculating the odds of matriculating at one of the schools that you are interviewing at.

Yeah I definitely don't buy this except for the case of true superstars. Schools evaluate students differently and are looking for different things. This, combined with the fact that most matriculants are only offered one acceptance do not offer much support to your argument. Like I said, for true superstars what you are saying is true, but these superstars are few and far between.

 

[Gurby]

I interpreted it as the probability of one person being accepted over the course of 3 IIs

So if we interpret it as a coin flip (p= 0.5) and the events are mutually exclusive, you sum the probabilities.

1/2 + 1/2 + 1/2 = 3/2 =150%

Am I misdefining the events in terms of independence?

3 coinflips would be, 50% chance of not getting in each time: 0.5 * 0.5 * 0.5 = 0.125 = 12.5% chance of not getting in = 87.5% chance of getting in. It's probably unlikely that you actually have 50% chance at each individual school - some will be higher and some will be lower depending on strength of your app compared to average matriculant at the school, and how well you interview.

Do the accepted numbers include the waitlist?

Yes, afaik, which means that the number of outright acceptances is quite a bit lower than whatever the listed % is

 

[RogueBanana]

TIL that as long as you post it to Reddit before SDN you can put up stuff that is normally behind paywalls...

All I saw was a post with numbers, I have no idea where they came from

 

[efle]

All I saw was a post with numbers, I have no idea where they came from

Oh I'm not accusing you of anything, I saw this post over there too.

I'm just wondering what the mod rationale is. You would never be allowed to post a spreadsheet of MSAR or US News stats here because that's not publicly available info. If you anonymously posted it in a reddit comment and then linked from SDN to that reddit comment though...

 

[779163]

Correct, but we're just talking about theoretical odds for arguments sake here

Yeah, I understand. I just feel that simplistic math like this is a bit disingenuous and misleading. Say, for instance, you just receive 1 interview at Case Western. It has a 49.4% post-II acceptance rate; however, the average person with 1 interview probably has a significantly lower chance of acceptance than a coin flip; interviews don't take place on a level playing field, and it is more than likely that the person with 3 additional interviews elsewhere (average # of interviews received by a matriculant is 4) has attributes that look more attractive to an admissions committee.

We really do not know how accurately the number 87.5% reflects the actual percentage of those with three interviews getting at least one acceptance. Although 3 is the median number of interviews that matriculants attend according to the MSQ, this number suffers from survivorship bias. How many people with three interviews got rejected? Is it truly 12.5%? I would not be surprised if it is significantly higher.

Schools evaluate students differently and are looking for different things.

This, I think, is another BIG reason that you can't resort to simplistic calculations like this, even if that 87.5% is accurate. The calculation assumes that you have an average shot at every school you interview at.

This, combined with the fact that most matriculants are only offered one acceptance do not offer much support to your argument. Like I said, for true superstars what you are saying is true, but these superstars are few and far between.

The median number of acceptances is 1, but the mean is 2 (divide total acceptances by enrollees). You should do the math with the assumption that every person holds two acceptances.

 

[RogueBanana]

Oh I'm not accusing you of anything, I saw this post over there too.

I'm just wondering what the mod rationale is. You would never be allowed to post a spreadsheet of MSAR or US News stats here because that's not publicly available info. If you anonymously posted it in a reddit comment and then linked from SDN to that reddit comment though...

The funny thing is the post I found it on was like a year old, It wasn't my post. But good to know there's a workaround lol

 

[787593]

3 coinflips would be, 50% chance of not getting in each time: 0.5 * 0.5 * 0.5 = 0.125 = 12.5% chance of not getting in = 87.5% chance of getting in. It's probably unlikely that you actually have 50% chance at each individual school - some will be higher and some will be lower depending on strength of your app compared to average matriculant at the school, and how well you interview.



Yes, afaik, which means that the number of outright acceptances is quite a bit lower than whatever the listed % is

Phew..

 

[libertyyne]

The statistical modeling assumes interview results are random. They are not. This is the reason you could have 10 interviews and not end up with an acceptance. Vs having one interview and end up with an acceptance. Yes having more interviews is better, and yes going to interviews at schools that accept a large portion of students post II is better. But after that your own interview performance and internal grading rubrics that we may not be privy to play a large role. To an outside observer it may seem random, but it is not.

 

[779163]

The statistical modeling assumes interview results are random. They are not. This is the reason you could have 10 interviews and not end up with an acceptance. Vs having one interview and end up with an acceptance. Yes having more interviews is better, and yes going to interviews at schools that accept a large portion of students post II is better. But after that your own interview performance and internal grading rubrics that we may not be privy to play a large role. To an outside observer it may seem random, but it is not.

This is why I advise against thinking about "odds" of success. Yes, there is some value in knowing the success rate of people who had similar circumstances as you did (e.g. the success rate of first-time applicants with your GPA/MCAT should inform your application strategy). But we do not know the rate of acceptance for people with 3 interviews, and we shouldn't pretend that 87.5% is that number based on rudimentary statistical models.

 

[RogueBanana]

Mods can we delete this thread? I don't want to share copyrighted info. It isn't fair to those who paid for it.