comparing means

#1
I have collected attention data from two different groups. for the first group (14 people), I measured attention (dependent variable) for 7 different attributes.
for the second group (17 people), I measured attention (dependent variable) for 27 different attributes (including those 7 attributes).
my question is which statistical analysis I should use to rank attention (dependent variable) of attributes within each sample?
also, which statistical analysis I should use to compare the attention of each attribute across two samples
Many thanks for your help
Nazila
 
#2
The sample size is too small for any standard non-parametric tests to be accurate. You should program a customized test. The test would be based an aggregate statistic, calculated off the ranks of the 7 attributes when all of them are pooled together (from both groups). The statistic would be a measure of the difference of ranks between the two groups. The distribution of the statistic under the null hypothesis could be approximated via bootstrap.

R and Matlab are the best programming languages for implementing the test.