Why 75/25 medians? why not means? Forum

[patentlybored]

Why are admission percentile statistics (25/75th gpa/lsat) reported in terms of medians rather than a single mean? Reporting scores as 25/75 medians gives schools an incentive to admit splitters and reverse splitters, whom i have nothing against, but nonetheless LS's are given the ability to impove their admission statitistics by accepting either type of splittter. Why not report admission stats as a straight-up mean?

[IAFG]

How would means disincentivize admitting splitters and reverse splitters?

[top30man]

How would means disincentivize admitting splitters and reverse splitters?

[cd7573]

Using means would provide a disadvantage to admitting super splitters Ex: 2.8 168's at WUSTL

[IAFG]

Using means would provide a disadvantage to admitting super splitters Ex: 2.8 168's at WUSTL

Means would be just as susceptible to manipulation, if not more. 25/75 at least requires that the middle 50% of admits are in a disclosed range.

[patentlybored]

Using medians gives schools no incentive to admit a 2.8/176 vs. a 0.0/176. Either way, the 25% gpa is the same..,

[rayiner]

Using means would provide a disadvantage to admitting super splitters Ex: 2.8 168's at WUSTL

LOL @ 2.8/168 being a "super" splitter.

[IAFG]

Using medians gives schools no incentive to admit a 2.8/176 vs. a 0.0/176. Either way, the 25% gpa is the same..,

Well first, the USNWR ranking is based on the 50th percentile median and not the 25/75, so it doesn't matter at all. That aside, means would allow schools to admit half 160/4.0s and half 180/3.0s and have a 170/3.5 mean.

[cd7573]

How many people get 180's?

[moonman157]

But really though, why means and not medians? With medians, a 2.5/174 is the same as a 3.5/174. With means they're not the same at all, and one candidate is clearly better than the other.

[patentlybored]

Using medians gives schools no incentive to admit a 2.8/176 vs. a 0.0/176. Either way, the 25% gpa is the same..,

Well first, the USNWR ranking is based on the 50th percentile median and not the 25/75, so it doesn't matter at all. That aside, means would allow schools to admit half 160/4.0s and half 180/3.0s and have a 170/3.5 mean.

Hypothetically this is true, bUT how many 180/3.0 ss's really exist? F*** ill come up with more intelligible logic tomorrow while briefly sober

[IAFG]

How many people get 180's?

How many people graduate with 0.0s?!

[birdlaw117]

With medians instead of means you at least know that at least half of people (or 25%, 75%, etc.) meet a certain threshold. A mean gives you no such real confidence in that fact. A school that admit 4 people with 175s and a 5th with a 150 would have a mean of 170, but an applicant with 170 doesn't appear to be in a very good position since 80% of students have a better LSAT.

Obviously with a sample size of 200+ this becomes pretty not important, but a data point is almost surely a better tool for making decisions in this context than a mean.

[cd7573]

How many people get 180's?

How many people graduate with 0.0s?!

Not my hypo... Try again

[cd7573]

With medians instead of means you at least know that at least half of people (or 25%, 75%, etc.) meet a certain threshold. A mean gives you no such real confidence in that fact. A school that admit 4 people with 175s and a 5th with a 150 would have a mean of 170, but an applicant with 170 doesn't appear to be in a very good position since 80% of students have a better LSAT.

Obviously with a sample size of 200+ this becomes pretty not important, but a data point is almost surely a better tool for making decisions in this context than a mean.

But the median here would be 175

[birdlaw117]

With medians instead of means you at least know that at least half of people (or 25%, 75%, etc.) meet a certain threshold. A mean gives you no such real confidence in that fact. A school that admit 4 people with 175s and a 5th with a 150 would have a mean of 170, but an applicant with 170 doesn't appear to be in a very good position since 80% of students have a better LSAT.

Obviously with a sample size of 200+ this becomes pretty not important, but a data point is almost surely a better tool for making decisions in this context than a mean.

But the median here would be 175

Yes. Yes it would. And without a 175 you are very unlikely to be admitted. I don't see the problem.

[IAFG]

How many people get 180's?

How many people graduate with 0.0s?!

Not my hypo... Try again

Go to hell.

[cd7573]

How many people get 180's?

How many people graduate with 0.0s?!

Not my hypo... Try again

Go to hell.

.

Sure thing bud

[rayiner]

What's the point of disincentivizing the admission of splitters? Does it make a material difference whether a school has a bunch of 2.9/175's versus 3.2/173's?

[IAFG]

But really though, why means and not medians? With medians, a 2.5/174 is the same as a 3.5/174. With means they're not the same at all, and one candidate is clearly better than the other.

Only if you believe that all 3.5s are better than all 2.5s. Which is obviously retarded.

[IAFG]

What's the point of disincentivizing the admission of splitters? Does it make a material difference whether a school has a bunch of 2.9/175's versus 3.2/173's?

The point is, people with 164/3.5s are mad as fuck

[rayiner]

What's the point of disincentivizing the admission of splitters? Does it make a material difference whether a school has a bunch of 2.9/175's versus 3.2/173's?

The point is, people with 164/3.5s are mad as fuck

As long as 170+ LSAT's remain rare, schools will eat the 2.x super-splitters to get those LSAT's.