How is LSAT score calculated?

[MatthewJBurns]

I guess this is going to be my inaugural post.

Sorry if this is a stupid question, but its been bothering me for awhile now. I've looked around quite a bit, but haven't been able to find definitive/clear answers on the subject.

I know my score is determined by comparing me against the people who wrote the test over the last three years, but does that mean my percentile score is determined in the same way and not exclusively tied to the test I'm taking?

Another question, does scoring a 160 always put the test-taker in a specific percentile (i.e. the amount of questions you need to score correctly to get a 160 changes relative to the difficulty of the test, but 160 always equals the 80 something percentile).

I've tried searching the forums, but the keywords I'm using are too broad and I can't really narrow it down.

There must have been an explanation written by someone in the long history of this forum. If someone could point me in the right direction it would really be appreciated. Again, I'm really just looking for the mechanics of how scores/percentiles are calculated.

Cheers,

[Julie Fern]

julie explain many times, but no one believe her.

[vesperholly]

Alpha-Score has a chart that suggests the scaled scores always correlate to percentiles (which makes sense): http://www.alpha-score.com/lsat-resources/lsat-scores-explained/

It's a good question.

[Julie Fern]

hey, you figure higher scaled score, higher percentile?

[!закон и право!]

Alpha-Score has a chart that suggests the scaled scores always correlate to percentiles (which makes sense): http://www.alpha-score.com/lsat-resources/lsat-scores-explained/

It's a good question.

Firstly, that chart is extremely outdated. Few if any LSAT administrations in the last few years have seen an 86 RAW yield a 170 scaled score.

Secondly, the percentiles for scaled scores do vary slightly (1-3 %ile points) by administrations. For example, a 155 scaled score was the 67th percentile a few years ago, but recently it has tended to fall into the 64th percentile. Likewise, the threshold for a 99th %ile score has increased by a scaled point or two in the last few years. This isn't a "natural" property of the test - rather, I suspect LSAC purposefully adjusted the score to percentile pairing to account for the greater frequency of 170+ scorers in the last several years.

To the OP:

A percentile rank is simply a statistical metric that is calculated by comparing your RAW score to the aggregate RAW scores of all test takers in the last 3 years. Your post seems to indicate that you have some idea what the percentile rank implies, so I am not really sure what you are asking.

The specific calculation for obtaining the percentile rank is: Cf + 0.5 (fi) / N * 100% where Cf is the cumulative frequency for all scores lower than the score of interest (i.e. all scores lower than your RAW score), fi is the frequency of the score of interest (i.e. your RAW score), and N is the number of examinees in the sample (i.e. all examinees in the last 3 years).

[vesperholly]

Thanks, Thales.

[MatthewJBurns]

Thanks all.

I've got a follow up.

So if my percentile score is calculated by comparing my raw score to that of those who wrote in the past three years and I happened to write the easiest test in that three year cycle, my 120-180 score will adjust accordingly (tough scale like -8 or -7 for 170) but my percentile could potentially be very high, correct?

Example:

On a harder test in the cycle, someone gets 75 questions right, scores 160 and is in the 80th percentile

On a much "easier" test in the cycle, someone gets 80 questions right, scores 160 and is in the 86th percentile.


I'm preoccupied by this because I'm applying to a Canadian school and they use your percentile score for admission. I wrote my LSAT in Sept 2009 and I honestly think the test was much easier than anything else I've seen in the previous three years and I've written them all. I'm predicting 170 will be -8 or so and while that will hurt my 120-180 score, if my reasoning is correct it shouldn't matter because I only benefit when you calculate the percentile score and that's what my target school considers.

Let me know if I'm way off here.

[compaq1984]

I dont remember seeing -8 as 170 for some time...good for you but just saying... -8 is usually 172+

[!закон и право!]

I dont remember seeing -8 as 170 for some time...good for you but just saying... -8 is usually 172+

IIRC, December 2004 (or 2005) was a -8 scale for 100 questions.

[!закон и право!]

Thanks all.

I've got a follow up.

So if my percentile score is calculated by comparing my raw score to that of those who wrote in the past three years and I happened to write the easiest test in that three year cycle, my 120-180 score will adjust accordingly (tough scale like -8 or -7 for 170) but my percentile could potentially be very high, correct?

Example:

On a harder test in the cycle, someone gets 75 questions right, scores 160 and is in the 80th percentile

On a much "easier" test in the cycle, someone gets 80 questions right, scores 160 and is in the 86th percentile.


I'm preoccupied by this because I'm applying to a Canadian school and they use your percentile score for admission. I wrote my LSAT in Sept 2009 and I honestly think the test was much easier than anything else I've seen in the previous three years and I've written them all. I'm predicting 170 will be -8 or so and while that will hurt my 120-180 score, if my reasoning is correct it shouldn't matter because I only benefit when you calculate the percentile score and that's what my target school considers.

Let me know if I'm way off here.

Personally, I suspect that the Sept 2009 scale will fall in line with recent test administrations (-10 or -11). I think it's unreasonable to suspect a -8 scale. There were swaths of people complaining about the difficulty of certain LR and LG questions.

Anyway, you're wrong about the percentile to scaled score calculations. A 160 has never been as high as the 86th percentile. Your percentile will depend entirely on the percentage of people falling below a given scaled score in the last 3 years (or 12 administrations). It has very little to nothing to do with how people performed on your given sitting.