help [Nonparametrics]
dear shri !
first I would suggest a descriptive summary table showing percentage of scores by treatment
rows: treatment (a,b,c,d)
columns: scores (0,1,2,3)
cells: row percentages (total per treatment = 100%)
in the case that you are interested in a score of 3 between treatments I suggest to define a new variable with values of 0 (score<>3) and 1 (score=3). you may then calculate the treatment effect (odds ratio + corresponding confidence interval) between two treatments regarding the risk of experiencing a score of 3 (via logistic regression to model the whole data set like: new variable = treatment).
hope this helps
martin
first I would suggest a descriptive summary table showing percentage of scores by treatment
rows: treatment (a,b,c,d)
columns: scores (0,1,2,3)
cells: row percentages (total per treatment = 100%)
in the case that you are interested in a score of 3 between treatments I suggest to define a new variable with values of 0 (score<>3) and 1 (score=3). you may then calculate the treatment effect (odds ratio + corresponding confidence interval) between two treatments regarding the risk of experiencing a score of 3 (via logistic regression to model the whole data set like: new variable = treatment).
hope this helps
martin
Complete thread:
- help shri 2008-08-20 11:06
- n x k contingency table? Jaime_R 2008-08-20 12:03
- help ElMaestro 2008-08-20 12:29
- help martin 2008-08-20 12:58
- help shri 2008-08-20 13:26
- helpmartin 2008-08-20 14:14
- help shri 2008-08-20 13:26
- Higher order cross-over with ordinal response? d_labes 2010-06-30 16:16