# Help please - in middle of brain meltdown!!

#### Lavina

##### New Member
I am putting together a my research proposal on Chinese Herbs (as you can already see, statistics and I do not have much in common!). I have plowed my way through a wonderful textbook, but am still struggling to decide upon the tests that I should use for my data! The pilot study will be as follows:

2 small treatment groups (n=10 in each group), one group receives chinese herbal formula, the other group will receive the placebo (participants will be allocated randomly).

Participants required to complete a previously verified and validated Quality of Life questionnaire at three distinct times (baseline, +1mth & +3mth). The questionnaire divides into 4 distinct domains, and the score for each is a mean of the questions within that answer marked from 0-8 (with no = 0 and highly yes = 8)

this is where i start to get confused! it looks as though i should be using mixed ANOVA analysis, but my sample sizes are so small that i cant meet the parametric assumptions, so surely should use non-parametric tests????

any help to point me in the right direction much appreciated! i am currently in a spin with definitions and tests and am no longer convinced that i even know how many independent variables i have!!!! :shakehead

E

#### elnaz

##### Guest
hello
you did two works , the first you have studied validity of your questionnaire,
it is related to this statement of you "Participants required to complete a previously verified and validated Quality of Life questionnaire at three distinct times (baseline, +1mth & +3mth). " but about your analysis, you should do Anova , but before it , you should test of normality with kolmogorov- spmirnov test , if your data were normal you should do Anova otherwise you can do cox -box transformation for normality or you can use nonparametric test that, for your data mannwhitney is suitable, of your writting i think ,you have only two independent groups , so in nonparametric test , mannwhitney is suitable test.