# How to Perform Statistical Testing to determine best student in class?

I have a dataset of marks of 6 students - A, B, C, D, E, F, G for 30 subjects each.

Subjects Student A Student B Student C ....
Subject 1 marks...
Subject 2
Subject 3
... 30 subjects

The marks are mostly around 50 to 70, rarely rising to 80 or 90.

I wish to determine :

• which students are significantly best/worst
• in which subject students perform better

I want to achieve these inferences using statistical testing, and validate the results with p-value.

Which statistical test should I use for this purpose? Should I be using t-test or ANOVA or some other test? How should I formulate the problem and format the data to use such a test?

I am new to this subject, so kindly pardon my naiveness. A simplified yet detailed help would be highly appreciated.

Thank You

• You need to define "best" and "worst". There are many definitions even with numeric scores, and the issue gets even thornier with ordinal values like letter grades. For example, is a student who gets straight B's better than one who gets half A's and half C's? What about a student who gets mostly A's but several F's? A simple mean of numeric scores would be a reasonable approach, but it's unclear how letter grades should be weighted. Jul 28 at 15:07
• The grades are not letter grades but numerical marks. mostly ranging between 50 to 70. with few cases going up to 80 or 90. To be honest, I am not sure how to define best or worst. I just wish to state from this dataset that a specific student is significantly better than others. Or marks of students in a particular subject is significantly better than other subjects. I am unable to formulate the problem to make such inferences from this dataset. Jul 28 at 15:11
• I agree with @NuclearHoagie here; it should be up to you to define best and worst, not us. There are subjective judgements involved and the details matter.
– mkt
Jul 28 at 16:22