Test of differences (count)

[yoyong555]

Hi everyone.

The scenario is this:

In a trial, a person visits the center thrice. During each visit, the number of medicines he receives is counted.

Example:

                                       Baseline         6 months      1 year

                Person 1             9                     2                    6

I want to know if there is a significant difference in the number of medicines he received across three visits. What statistical test do I use? And how should I program it?

Thank you very much.

Happy holidays to everyone.

[PaigeMiller]

Duplicate question, please reply in the other thread. https://communities.sas.com/t5/SAS-Programming/Test-of-differences-counts/td-p/613453

DO NOT REPLY HERE

[Kurt_Bremser]

I merged the two posts, and moved everything to Statistical Procedures.

[yoyong555]

Hi everyone.

The scenario is this:

In a trial, a person visits the center thrice. During each visit, the number of medicines he receives is counted.

Example:

                                       Baseline         6 months      1 year

                Person 1             9                     2                    6

I want to know if there is a significant difference in the number of medicines he received across three visits. What statistical test do I use? And how should I program it?

Thank you very much.

Happy holidays to everyone.

[PaigeMiller]

It is impossible to determine what statistical test to use, or even IF there is a statistical test in this case, without further information.

Is the variability from time period to time period, or across individuals (which doesn't exist as you specified a single person in your example)? Is baseline equivalent to six months or a year or longer? What is the exact (please be specific) hypothesis you want to test? What should we assume is the statistical distribution of this data (normal, Poisson, binomial, other)???

[yoyong555]

Thanks @PaigeMiller

[StatDave]

If there are multiple people and you want to compare the times, then you can arrange your data with one measurement per observation (row) in your data set, and fit a Poisson GEE model similar to what is shown in the example titled "Log-Linear Model for Count Data" in the GENMOD documentation.  That example has an offset which you don't appear to have in this case, so the OFFSET= option will not be needed.