COMLEX scores and normal distribution

[mikeh27]

---

Members don't see this ad.

I am certainly no statistics major or expert, but all of this nonsense talk about percentiles has me quite confused. Several people have posted about how because you know the standard deviation and the mean, then you can determine your percentile but how can this be since the scores are NOT a normal distribution. Determining that people within 1 standard deviation of the mean represent 68th percentile etc. applies ONLY to normal distributions. If this theory is so correct regarding COMLEX scores and the mean is 500 and the standard deviation is 79, then you are saying that one standard deviation above and one below represents 68 percent within 421-579. 34 percent from 500-579 and 34 percent from 421-500. And since 400 is passing, if you scored a 399 then you FAILED but your z score of -1.28 means you fall in the 10th percentile and since i have heard the pass rate for comlex is 95-97 percent, that doesn't add up...only 3-5 percent fail but you scored better than 10 percent and still failed...anyone that can explain how this makes sense...i am all ears.

 

[MDmiracle]

actually the national average is around 90%. so someone scoreing a 399 would fall into the lower 10% of scores and this should then make sense. the fact that so many people take the test should create a normal distribution or close to it.

 

[mikeh27]

I have seen in several places that the national pass rate is between 95 and 97 percent which would mean that someone who fails should have a score below the 3-5 percent range and not 10 like is calculated.

 

[MDmiracle]

here is what I used to make that assertion.
http://www.dmu.edu/com/faq/ (2004)

http://www.healthsciences.okstate.edu/communications/press_release/2005/2005_08_04_comlexscores.html (2005)

THough I did find it a bit odd that in this paper, the 1999 pass rate was 94%,
with a 514 mean and SD of 75. So apparently, the mean is not, or was not always 500. with these numbers a 399 would be 6th percentile. So the mean that we are using may not be completely accurate, and we do not know the pass rates for this year.

http://www.jaoa.org/cgi/reprint/100/4/238.pdf#search=""comlex level 1" passing rate"

 

[Idiopathic]

pass rate equals for first time test takers. repeat testers pass at about a 70% rate and this is where the difference in stats comes from.