I have a data set with marks of 430 study subjects. But it is not normally distributed. However, I want to divide the sample into four equal segments according to their scores. Is it ok to divide it using quartiles not doing log transformation.
I think as long as you divide according to percentiles (i.e. 75%), your division would be correct. It would be a problem if you divided according to mean +/- standard deviations.
The division of cases by quartile for marks (presumably on a 0-100 scale) would be exactly the same using original scores vs. any monotonic transformation (though, with log scores, you'd have to add a constant to avoid any cases of "0" value). So, carry on.