topic Re: Multiple Comparisons in a contingency table in Discussions

Multiple Comparisons in a contingency table

marcax — Sat, 09 Sep 2017 06:19:49 GMT

Hello.

Assuming I have a nominal outcome (0/1) as my Y variable.

I have 5 different years with a specific incidence of events occuring each year (x1,x2,x3,x4,x5), but also an overall incidence over the years: x*.

Creating a Fit x/y and I get a contingency table, indicating me a difference over the years.

How can I however identify which years are independently different to x* ?

How can I calculate an OR per year (x1,x2,x3,x4,x5) to x* ?

Many thanks for helping me out here ! Marc

Re: Multiple Comparisons in a contingency table

KarenC — Wed, 13 Sep 2017 18:39:47 GMT

How about Analysis of Means for Proportions? It is an option in the fit y by x when you your outcome has two levels.

Re: Multiple Comparisons in a contingency table

Dan_Obermiller — Mon, 18 Sep 2017 14:42:46 GMT

Correspondence Analysis from the Fit Y by X output may help you, too. It is not a formal statistical test, but the plot gives a nice graphical tool showing you the associations that exist in the table. Items that are not "close" to each other would be more "independent".