It has been a very long dry run since my last statistical evaluation so I was wondering is there someone on this forum whom has a fresher understanding of the subject willing to help me out. Essentially, I need to evaluate and report on a group of exam scores for a new testing procedure we are piloting. The data will be categorised as follows: Subject ID, gender, NESB (non native English, english speaker), Prior education (levels 4...8 as per AQTF), test result(s). A control group will consist of experienced staff members.

What would be the best approach and why? Since there are only two groups and each is independent, I was considering t-test for independent samples. However, if the control group fails I wouldn't need a test to conclude that the exam was poorly presented. On the other hand, I would like to know the relationships between the other variables. Eg. Wilcoxon rank test to see if level of previous education has a bearing on outcome(?). Would a Kruskal-wallis one-way ANOVA be a valid way of representing the rankings between the control and experimental group. You get my drift...


If you have any other suggestions I would like to hear from you and how you would interpret the results.


OC