Analysis of categorical diary data

[#1]

Hello,

I am using jmp 15.0 and have diary data to analyze. Patients had to record symptoms for 3 days (fictive example), either no symptom occurred (blank), or mild or moderate or severe symptoms or any combination of these categories occurred.

I would like to count the patients with always the most severe symptom occurred within those 3 days e.g. a patient with symptom severe on day 1, moderate on day 2 and mild on day 3 should be counted as patient with severe symptom.

I attach a sample file with the data structure for clarification and the outcome should be 

Patient A: blank (meaning no symptom but counted as case)

Patient B: severe 

Patient C: severe 

Patient severe 

Patient E: moderate 

Patient F: mild

Summary then is:

one patient with no symptoms

one patient with mild symptoms

one patient with moderate symptoms

three patients with severe symptoms

Is there an easy way to analyze this as I am not so familiar with scripting. I tried Analyze/ Consumer Research / Categorical/ Multiple Responses by ID and managed to get an overview on patients with any symptoms but not in the way as described above. May be there is an easy way to prepare the data to use this platform? The original data sheet I am intending to analyze has several thousands rows. 

Many thanks for your support!

[#2 · Accepted solution]

Most likely there is even easier way, but you could first recode/use formula to change those Character categories to numeric values:

Create summary table (or use tabulate) with Max of the new column:

Recode values back to categories:

Create summary/tabulate of that:

This whole process can be fairly easily scripted if needed.

[#3 · Accepted solution]

Many thanks the steps as described by you are working well. I am still on a beginner level with jmp and not there yet to script all that (my file has a lot of different symptoms in different rows) but I will look and get into it  step by step to manage that over time. For the time being - great that this works!