Exam Problem: calculating the overall distribution based on individual parts' distributions?

i40i

New member
Joined
May 13, 2022
Messages
1
Hello, hope you are enjoying the day!

So I am having this problem where I have a population of people that participated in a national examination process (N=20,500). The examination required each student to take 4 different tests A, B, C and D and then they would be graded on a scale of [0-20] for each test and an average of the four would form the final grade on a scale of 0 to 20. When the examination process was completed the competent autority uploaded the statistics showing what percentage of the students had scored in each grade category (0-5), (5-10), (10-11), (11-12), ..., (18-19) and (19-20) for every different test. No statistics where uploaded regarding the final score.

And here comes my question: I know in which grade bracket I belong for every of the four tests. Is it possible to calculate approximately in what percentile my overall score would be - based on the distributions of each of the four tests?
 
Hello, hope you are enjoying the day!

So I am having this problem where I have a population of people that participated in a national examination process (N=20,500). The examination required each student to take 4 different tests A, B, C and D and then they would be graded on a scale of [0-20] for each test and an average of the four would form the final grade on a scale of 0 to 20. When the examination process was completed the competent autority uploaded the statistics showing what percentage of the students had scored in each grade category (0-5), (5-10), (10-11), (11-12), ..., (18-19) and (19-20) for every different test. No statistics where uploaded regarding the final score.

And here comes my question: I know in which grade bracket I belong for every of the four tests. Is it possible to calculate approximately in what percentile my overall score would be - based on the distributions of each of the four tests?
The short answer is yes you can, but you'd have to make too many assumptions to be to "average" the aggregate distribution. Your result can be skewed and isn't precise.
 
Top