cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Try the Materials Informatics Toolkit, which is designed to easily handle SMILES data. This and other helpful add-ins are available in the JMP® Marketplace
Choose Language Hide Translation Bar
mmkaz
Level III

Non-Parametric-Test (JMP 17.2) Beginners Question

Hallo liebe Community,

ich habe eine ziemlich basale Statistik Frage:

Ich habe ein "Medikament A", das nehmen etwa 200 Patienten ein
Jetzt möchte ich die Gruppen vergleichen: 
Nebenwirkung B "Present" oder "Absent" 

Welchen Test verwende ich da? Ich dachte non-parametric über fit y to x und Wilcoxon-Test mit zwei Stichproben, Normal-Approxmation unter "Wahrscheinlichkeit > Z" den P-Value ablesen, um festzustellen ob der beobachtete Unterschied signifikant ist oder nicht.

Stimmt das denn so? Stehe etwas auf dem Schlauch gerade.

Vielen Dank!

4 REPLIES 4
mmkaz
Level III

Betreff: Non-Parametric-Test (JMP 17.2) Beginners Question

Kurzer Zusatz: Ich ermittle die Median Dosis bei je "Present" und "Absent" und vergleiche dann die Gruppe

 

MRB3855
Super User

Betreff: Non-Parametric-Test (JMP 17.2) Beginners Question

Hi @mmkaz : For clarity, does your data look something like the following or am I missing something?

MRB3855_0-1718190272995.png

 

mmkaz
Level III

Betreff: Non-Parametric-Test (JMP 17.2) Beginners Question

Genau so!

 

MRB3855
Super User

Betreff: Non-Parametric-Test (JMP 17.2) Beginners Question

Hi @mmkaz : OK. Using the Distribution platform, put Side Effect B in for Y as shown below.

 

MRB3855_0-1718205439702.png

Then click on the red triangle and choose choose "Confidence Interval" as shown below (I selected 95%).

MRB3855_1-1718205577517.png

You will the get a confidence interval for each (as shown below)

MRB3855_2-1718205657326.png

In this example case, 55% of this sample of 200 patients did not experience this side effect. The 95% CI is 48.1% to 61.7%. This basically means that in the entire population of patients (what you are really interested in), the percent who do not experience this side effect could be as low as 48.1% and as high as 61.7% (with 95% confidence).

 

You don't need to look at both Levels (since %Present = 100 - %Absent) . Pick the Level of interest and focus there. 

 

Questions? Come back,