PSY 520 Week 1: percentiles and exam forms

Be precise about what a percentile is, because the whole question turns on it. A percentile rank locates a score within a distribution: the sixtieth percentile means sixty percent of that group scored at or below it. It is a relative position within a reference group, not a measure of absolute performance. That immediately creates the problem in this scenario. If each form's percentiles are computed within the group that took that form, then a student at the fortieth percentile on Form B and one at the fortieth on Form A occupy the same relative position in different groups — which tells you nothing about whether the forms differed in difficulty. Notice too that the prompt does not say the student is wrong. The claim may be correct: forms genuinely differ unless work has been done to equate them. The question is what evidence would establish it.

So what can percentile data do here? It is genuinely useful if computed across all three groups pooled, because then a raw score's percentile in the combined distribution can be compared with its percentile within its own form, and a systematic gap is evidence. Comparing the raw score distributions directly is simpler still: means, medians and spread by form. But note the confound that makes this hard and is the sophisticated part of the answer — if students were not randomly assigned to forms, a lower mean on Form B could reflect a weaker group rather than a harder paper, and no amount of statistics separates those without randomisation or a common set of anchor items. Say that. The professionally correct answer is that establishing equivalence between forms is done in advance through test equating, not afterwards from one student's complaint — though the complaint is a reasonable prompt to check.