cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Choose Language Hide Translation Bar
rhammock
Level I

Trying to create a binary logistic regression that accounts for repeated measures

I am trying to create a logistic regression with a Yes/No response for a continuous variable. I followed the "Fit Y by X" menu on JMP to get the attached graph, but I am confused on why data points are falling between "0" and "1". There were only two possible categories. I would also like to make this logistic regression account for repeated measures since there were many independent replicates for some individuals. 

6 REPLIES 6
ron_horne
Super User (Alumni)

Re: Trying to create a binary logistic regression that accounts for repeated measures

Hi @rhammock ,

have a look at the 2nd example here.

please let us know if it helps

 

 

rhammock
Level I

Re: Trying to create a binary logistic regression that accounts for repeated measures

If I'm understanding correctly, I need to run my data through the generalized mixed model to account for the repeated measures and then do the logistic regression? My situation in more detail - I have a different number of independent replicates for different individuals. "avg_red_sta" is the same for an individual regardless of how many times its been used in an independent replicate, but "land?" can vary between replicates. I want to make sure that each individual has an equal weight in the distribution. I have a column in my data with a corresponding value to each "avg_red_sta" value. 

dale_lehman
Level VII

Re: Trying to create a binary logistic regression that accounts for repeated measures

More to the point of your question, try this from a few years back: Logistic Plot interpretation of points.  I still find the graph confusing, but coloring the points does help.

rhammock
Level I

Re: Trying to create a binary logistic regression that accounts for repeated measures

I'm a little confused on concept of jitter. I understand that JMP uses it to avoid clumps of points at "0" and "1" for a logistic curve, but it is not random? Every time I re-run the "Fit Y by X" the spread is the same. Any help in understanding that would be appreciated, otherwise I get why JMP does logistic regressions like this. 

 

Also, I'm having trouble finding the "coloring by Status" option

 

Thank you for the reply!

Re: Trying to create a binary logistic regression that accounts for repeated measures

The y ordinate is random, but the platform uses the same seed for the random generator. Users complained about the changing position of the markers when they ran the Logistic platform again.

dale_lehman
Level VII

Re: Trying to create a binary logistic regression that accounts for repeated measures

Under the red arrow above Rows in the data table, color or mark by column - Status.