small sample size

[furkan]

Hi everyone!
Nice to meet people do same things on the web
Here the question:
I have independent two group of patients
first group have 10 patients
second have 5 patients
can I compare their surgical corrections (advancement measured as milimeters) with ManWhitney test?
(since the sample size is small I cant use unpaired T test).
Can I assume the results with p<0.05 significant?
or should I assume p<0.01 signigicant?
Or essential question, Can I do statistic such a small number of patients?
I am not expert but I can use medcalc program and SPSS for simple tests.

thanks for answers

 

[trinker]

(since the sample size is small I cant use unpaired T test).

why not?

df = N-2
df = 15-2
df=13

It might be unlikely to detect a small difference but the test can be used on a sample size as small as N=4 where each group has > 1 observations.

Or essential question, Can I do statistic such a small number of patients?

The ttest was designed for small sample sizes (LINK)

 

[furkan]

Thanks
Then the other question coming;
Should I test distrubitions of the groups before I do ttest?
which test I should used to?
I know that if distrubition is normal I can do independent t test, otherwise I should do Man Whitney U test. Is it true?
thanks again for spending your time

why not?

df = N-2
df = 15-2
df=13

It might be unlikely to detect a small difference but the test can be used on a sample size as small as N=4 where each group has > 1 observations.


The ttest was designed for small sample sizes (LINK)