Difference between clinical significance and statistical significance

Question

I am working on a clinical data analysis project and am from IT domain.

While I am analyzing the data, I came to know that though some of the variables are not statistically significant, they are clinically significant.

I also read that difference in median values between case and control group for variables is used as a measure to indicate clinical significance

Can someone help me understand this with a simple example as to why difference in median between groups is considered for clinical significance?

Answer 1

Another viewpoint could be : statistically significant implies use of sampling theory to establish presence of significant difference between case and control groups. The "clinical " significance seems to test for correctness or validity of measures of variables. Apparenty, the measurement theory is meant for checking validity of data - the difference could arise on account of incorrect measurements. The non-parameteric tests - it seems - take care of measurement errors and hence, produce perhaps more reliable results.

Answer 2

This is non-parametric statistics, the mean requires confirmation to the normal distribution (but there are exceptions), or perhaps better put it requires a fixed relationship between mean and standard deviation. Non-parametric stats always use the median and are distribution free. Wilcoxon test/ Mann-Whitney U test are used to replace t-test. The Kruscal-Wallace test replaces the 1-way ANOVA Finally the Friedman test replaces 2-way ANOVA but has its critiques. The t- and ANOVA tests require normality, F-max is one approach to determine normality.