How to calculate your percentile rank among your ethnicity Forum

[marlo45]

Ok, so i've been wondering how to calculate my percentile among my ethnicity (Black) and thought i'd share what i found. LSAC released this report on score breakdown (it has a score breakdown by race/ethnicity that we will use): http://www.lsac.org/lsacresources/Resea ... -10-03.pdf

If you want to calculate your percentile among a select group, you need to find the race/ethnicity in the report, get the mean LSAT score and the standard deviation (SD) [for that group] for whatever year (i use the latest available; 2008-09). You need to calculate your z-score then use a z-score chart (this http://www.regentsprep.org/Regents/math ... ZChart.htm) to find your corresponding percentile.

Calculation example:
Chart for African Americans in 2008-09 says that...
Mean LSAT score = 142.25
SD = 8.5

z-score formula:
z-score = (your LSAT score - Mean LSAT score) / SD

z-score = (166 - 142.25)/8.5
z-score = 2.79

Now go to the z-chart and scroll vertically in the z column for 2.7 then horizontally for .09 (2.7+.09). The number in the box times 100 is your percentile among that group. For this example it's .9974 * 100 or 99.74th percentile. So it looks like i scored in the 93rd percentile overall but in the 99th percentile among my group.

Hope that helps someone because i couldn't find it here.

Edit: This can also be used to estimate the number of people among any of the groups that scored at or above a certain score. For the above example you could subtract the number in the box retrieved from the z-chart from 1 (this: 1 - .9974). Multiply the result by the total number of people in that group that took the test. In this case it would be .0026 * 13205 or roughly 34 or so that scored at or above 166.

[lovejopd]

Ok, so i've been wondering how to calculate my percentile among my ethnicity (Black) and thought i'd share what i found. LSAC released this report on score breakdown (it has a score breakdown by race/ethnicity that we will use): http://www.lsac.org/lsacresources/Resea ... -10-03.pdf

If you want to calculate your percentile among a select group, you need to find the race/ethnicity in the report, get the mean LSAT score and the standard deviation (SD) [for that group] for whatever year (i use the latest available; 2008-09). You need to calculate your z-score then use a z-score chart (this http://www.regentsprep.org/Regents/math ... ZChart.htm) to find your corresponding percentile.

Calculation example:
Chart for African Americans in 2008-09 says that...
Mean LSAT score = 142.25
SD = 8.5

z-score formula:

Thx, recap of my statistics class Z-Score!!
z-score = (your LSAT score - Mean LSAT score) / SD

z-score = (166 - 142.25)/8.5
z-score = 2.79

Now go to the z-chart and scroll vertically in the z column for 2.7 then horizontally for .09 (2.7+.09). The number in the box times 100 is your percentile among that group. For this example it's .9974 * 100 or 99.74th percentile. So it looks like i scored in the 93rd percentile overall but in the 99th percentile among my group.

Hope that helps someone because i couldn't find it here.

Edit: This can also be used to estimate the number of people among any of the groups that scored at or above a certain score. For the above example you could subtract the number in the box retrieved from the z-chart from 1 (this: 1 - .9974). Multiply the result by the total number of people in that group that took the test. In this case it would be .0026 * 13205 or roughly 34 or so that scored at or above 166.

Thx! I am glad that I saw 'z-score' on TLS~Recap of Statistics 101!

[Nova]

Very helpful!

I wonder if any one is ambitious enough to make a full chart from say 160- 99.9% for the URMs...

[2014]

Protip: If you want to do better on the LSAT don't denote your gender or ethnicity. Depending on what your gender or ethnicity actually is you are in for a gain of 2 to 15 points!

Correlation = Causation

[marlo45]

Protip: If you want to do better on the LSAT don't denote your gender or ethnicity. Depending on what your gender or ethnicity actually is you are in for a gain of 2 to 15 points!

Correlation = Causation

Not trying to be rude, but what exactly is your point? Are you suggesting that indicating one's gender/ethnicity could change the way his/her test is scored? If so, that's a conspiracy that could only exist in February since all other tests are released with answer data. If i'm wrong about your stance, please pardon my ignorance, but indeed, do clarify.

[2014]

It was sarcastic, clearly you don't get a boost for checking or not checking any particular demographic box. For some reason though people who did not denote their gender, their ethnicity, or both did significantly better. Obviously there is some other factor at play, but its interesting to see that the mean for those "private" individuals or w/e you want to call them is 3 or more points higher than those who denote their demographics.

[laxbrah420]

was that freakonomics or gladwell that found that?
those identifying as black males seem to lower their expectations for themselves and thus perform to that lower standard...

[marlo45]

It was sarcastic, clearly you don't get a boost for checking or not checking any particular demographic box. For some reason though people who did not denote their gender, their ethnicity, or both did significantly better. Obviously there is some other factor at play, but its interesting to see that the mean for those "private" individuals or w/e you want to call them is 3 or more points higher than those who denote their demographics.

Lol, now i get you. It's also interesting that 'Foreign Countries' score higher than the US.

[Its Always Sunny]

so essentially some ethnicity groups boost the curve. If there was an increase of applicants in African Americans AND their mean stayed 142 wouldn't that result in more generous curve due to an increase in lower scores?