How accurate is this ls admissions calculator? Forum

[tomrogers2010]

http://lawschoolcalculator.com

I used it and it appears to be rather reasonable.

[Knock]

From here: http://www.lawschoolpredictor.com/?page_id=173

The admit rate was calculated by taking the number of admitted applicants divided by total applicants for each respective prediciton category (total applicants includes admitted, waitlisted, and rejected applicants, but does not include those who status was listed as “pending”). The prediction categories used in testing accuracy are the same as LSP’s prediction engine with the exception of early prediction (which was not tested); predictions are adjusted to account for splitter-ness, weak GPAs, and URM status.

All Applicants

Target admit rates, based on how LSP renders predictions:
When LSP said Admit: >= 87%
When LSP said Strong Consider: ~ 69%
When LSP said Consider: ~ 50%
When LSP said Weak Consider: ~ 31%
When LSP said Deny: <= 13%

From my phone.

Edit: just noticed you were talking about a different site. I would recommend LSP anyways (assuming your not just advertising for your website, which you probably are).

[thecilent]

http://lawschoolcalculator.com

I used it and it appears to be rather reasonable.

Looks pretty legit. I like it

[tomrogers2010]

Edit: just noticed you were talking about a different site. I would recommend LSP anyways (assuming your not just advertising for your website, which you probably are).[/quote]

This is not my site. But I do have to say that it seems to way more accurate than LSP. LSP says that a URM with a 170 and 3.7 would be denied from Harvard and Yale and a "weak consider" for the next top four schools? Come on now. We all know better than that.

--LinkRemoved--

The above site seems to be more realistic and it definitely runs parallel with the stats of people who were admitted to certain schools (i.e. T-14).

[Knock]

Just checked it out. It seems pretty interesting. I guess I'll have to let TLS know how accurate the various calculators were in regards to my application cycle when I'm finished with it .

[tomrogers2010]

Just checked it out. It seems pretty interesting. I guess I'll have to let TLS know how accurate the various calculators were in regards to my application cycle when I'm finished with it .

That is usually the only way to tell lol

[KibblesAndVick]

I really like how it lists the schools based on probability of acceptance rather than by USNWR ranking. I had no idea North Carolina Central University would be more difficult to get into than Columbia

I'd be really interested to see a comparison between this model and lawschoolpredictor

[tomrogers2010]

I really like how it lists the schools based on probability of acceptance rather than by USNWR ranking. I had no idea North Carolina Central University would be more difficult to get into than Columbia

I'd be really interested to see a comparison between this model and lawschoolpredictor

That is definitely odd. North Carolina Central University should not be more difficult to get into than Columbia. I received lower numbers for Columbia.

[Hey-O]

Okay, I'll wear the dunce hat. The center column is the probability but what are the other columns?

[KibblesAndVick]

Okay, I'll wear the dunce hat. The center column is the probability but what are the other columns?

It's a confidence interval.

That might not be very helpful

Basically, because he only has a sample of the data (self reported and self selecting people from LSN) he/she is creating a range for the probability. Instead of saying "There's a 50% chance you'll be admitted into UVA" it's saying "The odds of you being admitted into UVA are between 40% and 60%".

That's oversimplified and leaves out a bunch of important stuff but maybe it will help you get the gist of it.

[tomrogers2010]

Okay, I'll wear the dunce hat. The center column is the probability but what are the other columns?

It's a confidence interval.

That might not be very helpful

Basically, because he only has a sample of the data (self reported and self selecting people from LSN) he/she is creating a range for the probability. Instead of saying "There's a 50% chance you'll be admitted into UVA" it's saying "The odds of you being admitted into UVA are between 40% and 60%".

That's oversimplified and leaves out a bunch of important stuff but maybe it will help you get the gist of it.

So you would somewhat "abide" by the middle column and nd leave the left and right columns out of the prediction?

[KibblesAndVick]

I wouldn't ignore the two outer columns. The point is that there's a lot of uncertainty. No one can absolutely say that you have an 43% chance of acceptance to a given school. If there are two candidates with identical LSATs and GPAs one might get in and the other might get rejected.

What this model is doing is taking everyone who has similar numbers to you (so if you have a 3.8 and a 170 it's taking all the data points that are close by) and calculating how many people get in and how many people get rejected. The problem is that you only have a sample of the data (the people who use LSN, not every single person who applied to the school). To try and address this problem he's created a range of probabilities called a confidence interval.

The confidence interval is probably biased because LSN is not a random sample but that's probably getting ahead of ourselves.

The point is that you should look at the range if you want a "ball park" estimate of your chances at a given school. If the range is between 75% and 95% you know that you have a great chance of getting in. If the range is between 15% and 25% then you know that school is a reach. The middle number is less telling than the range.

[094320]

..

[Tanicius]

There's no way I have a minimum 65% chance of snagging NYU with a 170/3.75... Also, how the hell have more people gotten Columbia with my numbers than Chicago?

[KibblesAndVick]

Let's say you wanted to create a model that would predict the weight of a bear based on the length of its footprints. So you go to the zoo and you weigh a bunch of bears and measure their feet. You look at the data and you see that there is a positive correlation between the size of their feet and how much they weight (this makes sense, right?, bigger feet means bigger bears, means heavier bears). With this information you can create a model that would predict how much a bear weights based on it's footprints.

So the next day you're in the woods and you find a bear footprint and you measure it. Based on your data you should be able to get a rough estimate of how much the bear weighs. It has a 10 inch paw print so it should weigh between 200 and 250 pounds based on similar bears we measured in the zoo.

Now there are a lot of problems with this. Maybe the bears in the zoo are different from the bears in the wild (maybe the people on LSN are different from most law school applicants). Maybe there's a different paw print to weight relationship for male bears and for female bears (maybe URMs have different results than nonURMs). And so on and so forth. The point is that you need to take the estimate with a grain of salt.

NOW. One problem with this model is that the data is bi-variate. How much a bear weighs is a continuous variable. It can range from basically 0 to whatever the heaviest bear weighs. Law school admissions are yes and no. Either you get in or you don't. Because of this the model has to give you a probability. It can't simply say "Yes" or "No" it has to say "75% chance of Yes" and "25% chance of No".

I don't know what the actual model is that this website is using but there are a lot of issues behind the statistics.

Sorry that this is too long. But maybe that will help someone...