I have a group of COPD patients who went to a specific clinic. I want to compare the number of exacerbations they had in the 6-months prior to and 6-months after attending their first clinic visit. The number of exacerbations has a negative binomial distribution.
Is it possible to do such an analysis on either Prism or SPSS?
Alternatively, will a Wilcoxon test suffice?
Thanks!
For details see the cites listed here: https://www.google.com/search?rlz=1C1CHBF_enUS874US874&ei=WqQWXsWBFZvWtAbkg6aoCw&q=Statistical+inference+with+negative+binomial+populations&oq=Statistical+inference+with+negative+binomial+populations&gs_l=psy-ab.12...0.0..47542...0.0..0.0.0.......0......gws-wiz.neWaxVzSbe8&ved=0ahUKEwjFls7qz_XmAhUbK80KHeSBCbUQ4dUDCAs
Best wishes, David Booth
One of the best link regarding comparisons/matching negative binomial distribution
https://stats.stackexchange.com/questions/49614/comparing-two-vectors-from-negative-binomial-distribution-in-r
Note that your data seems to be paired (you have data from the same patients, before and after their first visit.
I don't think that Prism can do this, and I don't have SPSS.But I can tell you that it is quite simple to do in R (www.r-project.org):
If "exacerbations" is a numerical vector with the observed numbers of exacerbations, "time" is a factor with levels before/after, and "patient" is a factor with patient IDs (in a paired design each ID will occur twice), you can fit an appropriate negative binomial model with:
model
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS