I have two groups of patients, a control (A) and a treatment (B), and data for patients' weight in the beginning and in the end of a period of treatment in each group. I need to check if the dynamic of the mean weight in B-group (treatment group) is significantly better than in A-group (control group). Which of the specified two tests is appropriate? There's Kruskal-Wallis test to compare the difference between median of two independent samples (e.g. the median of weight in A and B in the end of the period) and t-test to compare to related samples (e.g. mean weight for B before and after the treatment and the same thing for A). But, it seems that it does not help me to prove the significance of difference between the dynamic of two groups.
$\begingroup$
$\endgroup$
3
-
$\begingroup$ what do you mean by dynamic of mean weight ? $\endgroup$– Subhash C. DavarFeb 25 at 17:06
-
$\begingroup$ yes, I summarize all weight for each prediod and then divide the sum of wights for the second period by the sum of the first one $\endgroup$– VladFeb 26 at 10:31
-
$\begingroup$ do you want a test for difference between the two ratios - ratio between before and after for treatment group and control group ? $\endgroup$– Subhash C. DavarFeb 28 at 14:26
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
Significance of difference between the dynamic (as defined in comment)of two groups can be ascertained by computing t-statistic of the difference in overall mean weight (before and after) for control group and overall mean weight (before and after) for treatment group. Compare the computed t-statistic with table value at a given significance level (say, .05). If computed t-statistic exceeds table value, there is effect of intervention on patients weight. The treatment group will have a lower weight after treatment than before the treatment.