Community Member

Joined:

Jun 28, 2016

How do I conduct a z test for multiple independent proportions

I have three independent samples and want to test if P1=P2=P3. I am attaching the file I have created in JMP. I can do this hypothesis testing using an online calculator: Z-test for Proportions-Independent Groups | McCallum Layton, but I would prefer to do this in JMP so that I can get the details. Can someone walk me how to do this?

Super User

Joined:

Feb 10, 2013

Re: How do I conduct a z test for multiple independent proportions

One option is to use Analysis of Means for Proportions (http://www.jmp.com/support/help/Analysis_of_Means_for_Proportions.shtml#121733).

Copy this script to a new script window, then with the table open that you shared run this script.  Then from the main red triangle: Script > Relaunch Analysis to see how I completed the dialog box. The ANOM for proportions is an option in the contingency platform. You can also change the ANOM plot to be for your "Yes" variable if you prefer (red triangle menu once you run the ANOM).  Note, this compares all three of your groups to the overall proportion.

Contingency(

Y( :Name( "Completed?" ) ),

X( :Name( "Labs?" ) ),

Weight( :Completed Counts ),

Contingency Table,

Analysis of Means for Proportions( 1, Point Options( "Show Needles" ) ),

SendToReport(

Dispatch(

{"Mosaic Plot"},

"Conting Mosaic",

FrameBox,

{DispatchSeg( MosaicSeg( 1 ), Cell Labeling( "Show Percents" ) )}

),

Dispatch(

{"Mosaic Plot"},

"Conting Mosaic Single",

FrameBox,

{Frame Size( 61, 240 ), DispatchSeg(

MosaicSeg( 1 ),

Cell Labeling( "Show Percents" )

)}

)

)

);