Dear subscribers of the SPM List,
I have a vector with parameter estimates (betas) from within a ROI,
extracted from different subjects. I would like to compute the
confidence intervals for this group. However, when looking at the data,
the histogram tells me they are not normally distributed. Furthermore,
both the Kolmogorov-Smirnov test and the Shapiro-Wilk test hint at
non-normally distributed data. These are null hypothesis significance
tests, so one should be weary of their interpretation. Nonetheless, the
more I look at the data, the less normally distributed they are :).
Can I assume normality anyway, due to the central limit theorem (CLT)?
Unfortunately, I am not able to sample 300 more subjects so that the CLT
can come into play. Or am I completely off here? I have to admit, my
statistics knowledge is somewhat limited.
What is the best approach here? Should I use bootstrapping? Or is the
course of action I want to take not particularly suited for this approach?
Regards,
Glad
--
Glad MIHAI, M.Sc. Biomedical Physics
Functional Imaging | University Clinic Greifswald
Walther-Rathenau-Straße 46 | 17475 Greifswald | Germany
Tel: +49 3834 86 69 44 | Fax: +49 3834 86 68 98
www.baltic-imaging-center.de
|