1
$\begingroup$
Heart                  Group-1                         Group-2
               Case1    Case2   Case3          Case1    Case2   Case3
İnflammation    -       -       -              +++      ++      ++
Hemorrhage      +       -       -              ++       +++     ++
Fibrosis        +       +       -              ++       +++     ++
Necrosis        ++      ++      +              ++       +++     ++

(-;absent, +;mild, ++;moderate, +++;severe)

A sample table is seen above. I have 3 cases in each groups. How can I calculate statistical significance between two groups. Should I calculate statistical significance for each lesion?

$\endgroup$
2
$\begingroup$

You have very small N's here. Some tests of statistical significance might be forced to give you a p-value that you want but they really aren't terribly valuable. You're best off describing your two groups and pointing out that, other than necrosis, there is no overlap between the two groups. No matter what a test said you'd really need to go get more data to get any kind of proper effect size measurement.

How are mild, moderate, and severe determined anyway? It seems to me that several of those response variables can easily have numerical measurements.

$\endgroup$
  • $\begingroup$ Thanks for answer John. I give only a sample table but i have 3 more that there is overlap. We would like to explain statistically these results, is it possible?. If scoring be numerical, is it useful for statistical analysis? Anyway i want to learn which statistical test should i use. If you suggest describing and pointing out although adversity of interpretation some tables, i will try to do this. $\endgroup$ – volkan Mar 1 '14 at 23:22
  • $\begingroup$ At least 6 is getting into a barely useful number of cases. But if there are numbers most definitely they are important. It can completely change the answer to the question. You should really the edit the question with much more information. $\endgroup$ – John Mar 2 '14 at 0:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.