does your logistic regression model suck?
DESCRIPTION
Does your logistic regression model suck?. PERFECTION!. This is bad. Model Convergence Status Quasi-complete separation of data points detected. Warning: The maximum likelihood estimate may not exist. Warning: - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/1.jpg)
Does your logistic regression model suck?
![Page 2: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/2.jpg)
PERFECTION!
![Page 3: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/3.jpg)
![Page 4: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/4.jpg)
![Page 5: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/5.jpg)
![Page 6: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/6.jpg)
This is bad
Model Convergence StatusQuasi-complete separation of data points detected.
Warning:The maximum likelihood estimate may not exist. Warning:The LOGISTIC procedure continues in spite of the above
warning. Results shown are based on the last maximum likelihood iteration. Validity of the model fit is questionable.
![Page 7: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/7.jpg)
Complete separation
X Group
0 0
1 0
2 0
3 0
4 1
5 1
6 1
7 1
![Page 8: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/8.jpg)
If you don’t go to church you will never die
![Page 9: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/9.jpg)
Quasi-complete separation
Like complete separation BUT one or more points where the points have both values
1 12 13 14 14 05 06 0
![Page 10: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/10.jpg)
there is not a unique maximum likelihood estimate
![Page 11: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/11.jpg)
“FOR ANY DICHOTOMOUS INDEPENDENT VARIABLE IN A LOGISTIC REGRESSION, IF THERE IS A ZERO IN THE 2 X 2 TABLE FORMED BY THAT VARIABLE AND THE DEPENDENT VARIABLE, THE ML ESTIMATE FOR THE REGRESSION COEFFICIENT DOES NOT EXIST.”
Depressing words from Paul Allison
![Page 12: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/12.jpg)
What the hell happened?
![Page 13: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/13.jpg)
Solution?
• Collect more data. • Figure out why your data are missing and fix
that. • Delete the category that has the zero cell..• Delete the variable that is causing the problem
![Page 14: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/14.jpg)
Model Fit Statistics
![Page 15: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/15.jpg)
![Page 16: Does your logistic regression model suck?](https://reader035.vdocuments.us/reader035/viewer/2022071807/56812cf2550346895d91c02b/html5/thumbnails/16.jpg)
Bonus fact!
• You can compare nested models using the model fit statistics and test if one model is superior to the other