Chi-Square Analysis: WHERE is it in JMP? Pretty simple in other software, but I would like to do it in JMP. I have data to determine if certain departments have disproportionately high injury rates compared to the overall rate. A goodness-of-fit exercise.
Solved! Go to Solution.
Here's what I am trying to get JMP to do for me (done in Excel): Calculated the Chi-Square Stats for each department and total:
|Dept||Staff Level||Observed Injuries||Staff Proportion||Expected Injuries||Expected-Observed||(Expected-Obs)^2||Chi-Squared|
|Totals||424||61||1.000||61||Total Chi-Sq =||25.26|
I believe Lou has the proper approach. However, his example included the full raw data table for the different departments. Just in case you had summarized data (as presented in the sample data you provided), you would want to conduct your analysis using Frequency variables. I have included that example here.
Dan and Lou,
Thanks for your help. I was having difficult switching the method of data entry from the way I was doing with another software package.
No problem....Thank you for being so open to share the data. It was an excellent opportunity to refresh my understanding of JMP for your application.
I actually think the format of the summary data in Steven's Excel spreadsheet is not that unusual. And I don't think it is always realistic to change the data entry format just to suit the idiosyncrasies of a particular analysis method. Which leaves (imho) contingency analysis languishing in Fit Y By X as the poor cousin to Bivariate and Oneway. It's tempting (especially for me) to write an add-in to deal with this problem, but it seems so basic and central to data analysis that JMP should deal with this. We have very powerful tools such as STACK and SPLIT that allow common data structures to be reshaped for particular methods of analysis. Is there not scope for a tool for reshaping of frequency-type data e.g. from Steve's format to DanO's ?