Is it possible to figure out your percentile (roughly)?

[automan2]

---

Members don't see this ad.

So, if we have a normal distribution, and we have a mean score of 218 and a standard deviation of 23. Are we able to estimate our percentiles with this data?

So 68% of the scores fall between 195 and 241.

That means your pecentile if you got a 195 would be about 16%.
Your percentile if you scored a 241 would be about 84%.

Is there a way to estimate percentiles? If this is a normal curve, you would think you could estimate were people fall.

Thanks for any help.

 

[Scaredshizzles]

Not a normal curve, so no.

If I had to guess, 185 is 6th percentile, 218 is about 52%tile, 240 is around 85th percentile, 250 is around 95-96th percentile, 260 is 99th percentile. 270 is like 99.7%ile, and so on.

 

[The Buff]

The curve may not be normal, but if we assume it is, you can just take your z-score [(your score-218)/23] and then translate it into a percentile. This website has a conversion table for z-scores to percentiles. http://www.acposb.on.ca/conversion.htm

At least it will be a rough estimate.

 

[automan2]

Cool, I just found this website as well, it correlates well with the z-score method above.


http://davidmlane.com/hyperstat/z_table.html

Go to the second calculator. Enter 218 as the mean, 23 as the standard
deviation. Select "Below." In the shaded area, enter a %ILE as a
fraction of 1 to determine a corresponding USMLE score. For example, to
determine the USMLE score that cooresponds to the 99%ILE, enter .99.
Beside "Below," your answer is 271.5.

 

[Miami_med]

Were it a normal curve,

Approximately
172=2.5%, 195=16%, 206=30%, 218=50%, 230=70%, 241=84%, 264=97.5%

This puts our 272 at the 99th percentile

 

[PeepshowJohnny]

I thought I remember hearing that the distribution of Step 1 was very much not a normal curve and that it was really impossible to calculate your percentile...

 

[nowai]

It would probably be slightly bimodal right?

 

[Scaredshizzles]

If you must know, find an old percentile chart given by NBME, construct a graph, and it would probably look similar, just with different numbers.And no it isn't a normal distribution. You can calculate SDs on non-mormalized distributions, the only difference is that it doesn't give you anywhere near accurate information about percentiles. If you took the average between the 16th and 50th percentile, and the 50th and 84th percentile, it would probably be close to the 23 SD they list (i.e. -1 SD to 1 S.D.) .....that's about as far as you can go where it would even be close.