Re: Percentile Scores

[Alan Mead]

When I was a young analyst, someone asked if the percentiles I'd provided were calculated the "right way" or the "wrong way." Using the cumulative percents is the "wrong way" (obviously, for some values of "wrong").

The "right way" (for some values of "right" or "correct") calculates the percentile of the midpoint of the score category and will never be exactly 0 nor exactly 100.

I calculate percentiles that way. I take the values of the frequency table and copy them into a spreadsheet.

percentile = ( 0.5 * COUNT + BELOW ) / TOTAL

where COUNT is the frequency of the current score category, BELOW is the cumulative count of cases with lower scores, and TOTAL is the total number of cases.

BTW, the other problem that arises is that some scores will be missing. This is less an issue for OP who just wants percentiles for his participants/students. One solution is to a single "pseudo count" person to ever score category (so it's COUNT+1 and BELOW and TOTAL include these "pseudo" people). I think this is Laplace smoothing.

-Alan

On 3/22/2023 10:18 AM, Peg Whalen wrote:

I believe you could just use the Cumulative Percent off the Frequencies table

On Wed, Mar 22, 2023 at 10:13 AM Ian McKenzie <Ian@prosocialscience.com> wrote:

Hi all!

Let's say that I have a database of scores by individual:  for example,
Sam has a score of 29 on his math test.  I have 140 cases that are
simple like that.

I would like to compute the percentile score for each individual such
that a case would look like this;


Name    Score    Percentile

Sam        29           39

...

I'd like the same kind of information for all 140 cases.

Anybody know how to do this?  I would truly appreciate a solution.

Thanks


Ian


--
This email has been checked for viruses by AVG antivirus software.
www.avg.com

-- 

Alan D. Mead, Ph.D.
President, Talent Algorithms Inc.

science + technology = better workers

https://talalg.com


Clarke's first law: When a distinguished but elderly scientist states
that something is possible, she is almost certainly right. When she
states that something is impossible, she is very probably wrong.