what statistical test can i use for my research ?


New Member
Hello everyone,
I'm currently in France for my Master degree of psychology. I work on a research about the question of the relationship between symptoms of conduct disorder and childhood depression through the representation of child attachment.

I have three variables:
-Dependent variable: depression rate for children measured with MDI-C test and CBCL (Child behavior Checklist)
-Independant variable : attachment type measured with the Attachment Story Completion Task (ASCT),(security style, inhibition of the attachment system, Hyperactivation of the attachment system, Disorganization)
-Independant variable: conduct disorder mesured with the CBCL (2 modes: présent, Absent)

My hypothesis are:
HO1 → The level of depression will be higher for children with conduct disorder than in children without conduct disorder.
HO2 → The level of depression will be higher for children with a fragile attachment link, unlike children with a secure attachment.
HO3 → The level of depression may be high for children with conduct disorder, especially if the child has an insecure/ disorganized attachment .

This study is conducted on a single group of 9 children, they will all go through the same tests. What kind of statistical analysis would you advise me to use for studying these hypotheses ??

Thanks you in advance for your help !


Well-Known Member
Hi Jon,
Generally sound like 3 two-sample t-tests, BUT the sample is too small for any preliminary assumption (normality, symmetry, no outliers )
You can try the Mann-Whitney U test, but I suspect it might be also too small ( for example compare group of 2 to group of 7 ...)
Also when doing several tests on the same population you should take a smaller significant level (α).