237046_en-US

liliconcom

immediate help

I require help in finding statistical tests to run for my survey research. My responses consist of nominal and ordinal variables but I only get mosaic plots and nothing concerning t-tests or Mann-Whitney U tests. I am not sure what I am doing wrong. "9" represents missing data. Any help would be appreicated

Re: immediate help

You can specify to JMP for the columns that contain "9"s as mssing values to set the Column Property called "Missing Value Codes" to recognize "9" as missing.

JMP does not calculate a Mann-Whitney U.  It uses Wilconon two group test, which is exactly the same as the specified Mann-Whitney U.  However,  the measurement data for that test requires numerical data, not categorical.

What I believe you need to do, is to read up on the Categorical Platform in the Consumer Research section of the Analyze pull down menu.  You can read the documentation under Help pull down menu

     Help==>JMP Documentation Library

Others on the Discussion Forum are more experienced with this area, and will probably have more input for you.

Re: immediate help

Jim,

Thank you for your assistance. I just wanted to ask two more questions: 

1) Based on the present information with missing data accounted for what would be the most appropriate test to identify possible intersubjectivity between responses?

2) By converting the responses present in numerical data and condensing the data table to a smaller size. Would this allow for the possiblity of approval for the use of t-tests, anova, and the  Wilconon two group tests? 

I am sorry for the intrusion, I was just looking for a second opinion in order to achieve workable results.