Tutorials: Tutorial on Count Regression

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

Atkins, D. C., Baldwin, S., Zheng, C., Gallop, R. J., & Neighbors, C. (in press). A tutorial on count regression and zero-altered count models for longitudinal addictions data.

本帖隐藏的内容

[Paper]
[RAPI Data]
[TLFB Data]
[README]
[R code]
[SPSS code]
[SAS code]
[Mplus code]
[Stata code]

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏2 回帖

关键词：regression Tutorials regressio Tutorial regress count

1. GENLINMIXED
   
2. /DATA_STRUCTURE SUBJECTS=id
   
3. /FIELDS TARGET=rapi
   
4. /TARGET_OPTIONS DISTRIBUTION=POISSON LINK=LOG
   
5. /FIXED EFFECTS=gender time gender*time
   
6. /RANDOM EFFECTS=time USE_INTERCEPT=TRUE SUBJECTS=id
   
7. COVARIANCE_TYPE=UNSTRUCTURED .
   
8. GENLINMIXED
   
9. /DATA_STRUCTURE SUBJECTS=id
   
10. /FIELDS TARGET=rapi
    
11. /TARGET_OPTIONS DISTRIBUTION=NEGATIVE_BINOMIAL LINK=LOG
    
12. /FIXED EFFECTS=gender time gender*time
    
13. /RANDOM USE_INTERCEPT=TRUE SUBJECTS=id COVARIANCE_TYPE=UNSTRUCTURED .

复制代码

1. Title:
   
2.     GLMM NB RAPI Hurdle
   
3. 
   
4. Data:
   
5.     File IS RAPI.final.csv;
   
6. 
   
7. Variable:
   
8.     Names ARE id rapi male time;
   
9.     Usevariables ARE id rapi male time;
   
10.     Cluster IS id;
    
11.     Count IS rapi (nbh);
    
12.     Within IS time;
    
13.     Between IS male;
    
14. 
    
15. Analysis:
    
16.     Type = twolevel random;
    
17.     Processors = 4;
    
18.     Integration = MONTECARLO;
    
19.     Estimator = MLR;
    
20. 
    
21. Model:
    
22.    
    
23.     %WITHIN%
    
24.     tslope | rapi ON time;         !fixed effect for time;
    
25.                                                         !tslope is the random effect for time;
    
26.     tslopei | rapi#1 ON time;        !fixed effect for time -- zeros;
    
27.                                                                 !tslope is the random effect for time -- zeros;
    
28.    
    
29.     %BETWEEN%
    
30.     rapi;                                         !random intercept;
    
31.     rapi#1;                                        !random intercept -- zeros;
    
32.     tslope rapi ON male;         !fixed effect for male and male x time;
    
33.     tslopei rapi#1 ON male;        !fixed effect for male and male x time -- zeros;
    
34.     tslope WITH rapi;                 !covariance between random effects;
    
35.     tslopei WITH rapi#1;        !covariance between random effects -- zeros;
    
36. 
    
37.    
    
38.    

复制代码

1. Title:
   
2.     GLMM Poisson RAPI Model 1
   
3. 
   
4. Data:
   
5.     File IS RAPI.final.csv;
   
6. 
   
7. Variable:
   
8.     Names ARE id rapi male time;
   
9.     Usevariables ARE id rapi male time;
   
10.     Cluster IS id;
    
11.     Count IS rapi;
    
12.     Within IS time;
    
13.     Between IS male;
    
14. 
    
15. Analysis:
    
16.     Type = twolevel random;
    
17.     Processors = 4;
    
18.     Estimator = MLR;
    
19. 
    
20. Model:
    
21.    
    
22.     %WITHIN%
    
23.     tslope | rapi ON time;         !fixed effect for time;
    
24.                                                     !tslope is the random effect for time;
    
25.     %BETWEEN%
    
26.     rapi;                                         !random intercept;
    
27.     tslope;                                 !random slope;
    
28.     rapi ON male;                         !fixed effect for male;
    
29.     tslope WITH rapi;                 !covariance between random effects;

复制代码

1. Title:
   
2.     GLMM Poisson RAPI Model 2
   
3. 
   
4. Data:
   
5.     File IS RAPI.final.csv;
   
6. 
   
7. Variable:
   
8.     Names ARE id rapi male time;
   
9.     Usevariables ARE id rapi male time;
   
10.     Cluster IS id;
    
11.     Count IS rapi;
    
12.     Within IS time;
    
13.     Between IS male;
    
14. 
    
15. Analysis:
    
16.     Type = twolevel random;
    
17.     Processors = 4;
    
18.     Estimator = MLR;
    
19. 
    
20. Model:
    
21.    
    
22.     %WITHIN%
    
23.     tslope | rapi ON time;         !fixed effect for time;
    
24.                                                     !tslope is the random effect for time;
    
25.     %BETWEEN%
    
26.     rapi;                                         !random intercept;
    
27.     tslope rapi ON male;         !fixed effect for male and male by time interaction;
    
28.     tslope WITH rapi;                 !covariance between random effects;
    
29. 
    
30.    
    
31.    

复制代码

1. Title:
   
2.     NB TLFB Hurdle;
   
3. 
   
4. Data:
   
5.     File IS TLFB.final.csv;
   
6. 
   
7. Variable:
   
8.     Names ARE id day drinks male weekend greek;
   
9.     Usevariables ARE id drinks male weekend greek mgreek;
   
10.     Cluster IS id;
    
11.     Count IS drinks (nbh);
    
12.     Within IS weekend;
    
13.     Between IS male greek mgreek;
    
14. 
    
15. Define:
    
16.         mgreek = male*greek;
    
17. 
    
18. Analysis:
    
19.     Type = twolevel random;
    
20.     Processors = 4;
    
21.     Integration = MONTECARLO;
    
22.     Estimator = MLR;
    
23. 
    
24. Model:
    
25. 
    
26.    
    
27.         %WITHIN%
    
28.         wslope | drinks ON weekend; !fixed effect for weekend;
    
29.                                     !wslope is the random slope for time;
    
30.         wslopei | drinks#1 ON weekend; !fixed effect for weekend - zeros;
    
31.                                        !wslope is the random slope for time - zeros;
    
32.         %BETWEEN%
    
33.         drinks; !random intercept;
    
34.         drinks#1; !random intercept -- zeros;
    
35.         wslope drinks ON male greek; !fixed effect for male, greek, ;
    
36.                                          !male x weekend and greek x weekend;
    
37.         drinks ON mgreek;            ! fixed effect for male x greek interaction;
    
38.         wslopei drinks#1 ON male greek; !fixed effect for male, greek, -- zeros ;
    
39.                                         !male x weekend and greek x weekend -- zeros;
    
40.         drinks#1 ON mgreek;             ! fixed effect for male x greek interaction -- zeros;
    
41.         wslope WITH drinks;                                ! covariance between random effects;
    
42.         wslopei WITH drinks#1;                        ! covariance between random effects -- zeros;
    
43.         

复制代码

1. Title:
   
2.     NB TLFB;
   
3. 
   
4. Data:
   
5.     File IS TLFB.final.csv;
   
6. 
   
7. Variable:
   
8.     Names ARE id day drinks male weekend greek;
   
9.     Usevariables ARE id drinks male weekend greek mgreek;
   
10.     Cluster IS id;
    
11.     Count IS drinks (nb);
    
12.     Within IS weekend;
    
13.     Between IS male greek mgreek;
    
14. 
    
15. Define:
    
16.         mgreek = male*greek;
    
17. 
    
18. Analysis:
    
19.     Type = twolevel random;
    
20.     Processors = 4;
    
21.     Estimator = MLR;
    
22. 
    
23. Model:
    
24.    
    
25.         %WITHIN%
    
26.         wslope | drinks ON weekend; !fixed effect for weekend;
    
27.                                                                 !wslope is the random slope for time;
    
28.         %BETWEEN%
    
29.         drinks;                                                !random intercept
    
30.         wslope drinks ON male greek; !fixed effect for male, greek, ;
    
31.                                                                  !male x weekend and greek x weekend;
    
32.         drinks ON mgreek;                         ! fixed effect for male x greek interaction;
    
33.         wslope WITH drinks;                        ! covariance between random effects;
    
34.         
    
35.         

复制代码

1. **************************************************************
   
2. ** Analyses of RAPI data for count regression tutorial *******
   
3. ** IR, 5/8/12                                                                                         ******
   
4. **************************************************************
   
5. 
   
6. /* read in file */
   
7. insheet using "http://depts.washington.edu/cshrb/newweb/stats%20documents/RAPI.Final.csv"
   
8. 
   
9. ************************************************************
   
10. /* Poisson mixed effects model with random intercept only */
    
11. ************************************************************
    
12. xi: xtmepoisson rapi i.gender*time || id:,
    
13. 
    
14. /* store model estimates for comparison to other models */
    
15. estimates store model0
    
16. 
    
17. **************************************************************************
    
18. /* Poisson mixed effects model with random effects for intercept + time */
    
19. **************************************************************************
    
20. 
    
21. xi: xtmepoisson rapi i.gender*time || id: time,
    
22. estimates store model1
    
23. 
    
24. 
    
25. 
    
26. ****************************
    
27. /* over-dispersed Poisson */
    
28. ****************************
    
29. 
    
30. /* create variable for over-dispersion term */
    
31. gen obs = _n
    
32. 
    
33. xi: xtmepoisson rapi i.gender*time || id: || obs:,
    
34. estimates store model2
    
35. 
    
36. ********************************************************
    
37. /* over-dispersed Poisson with random effect for time */
    
38. ********************************************************
    
39. 
    
40. xi: xtmepoisson rapi i.gender*time || id: time || obs:,
    
41. estimates store model3
    
42. 
    
43. ************************************************************************************
    
44. /* LR test comparing overdispersed models with and without random effect for time */
    
45. ************************************************************************************
    
46. 
    
47. lrtest model2 model3
    
48. 
    
49. **********************************************************************
    
50. /* LR test comparing non-overdispersed model to overdispersed model */
    
51. /* with random effects of intercept and time*/
    
52. *********************************************************************
    
53. 
    
54. lrtest model1 model3
    
55. 
    
56. 
    
57. 
    
58. ****************************************
    
59. /* Random intercept negative binomial */
    
60. ****************************************
    
61. xi: xtnbreg rapi i.gender*time, i(id) re /* re is random effects estimator which is already the default */
    
62. 
    
63. xi: xtnbreg rapi i.gender*time, i(id) irr  /* show rate ratios */
    
64. 
    
65. **********************************************
    
66. /* random intercept Poisson using xtpoisson */
    
67. **********************************************
    
68. xtset id time
    
69. 
    
70. xi: xtpoisson rapi i.gender*time, re
    
71. xi: xtpoisson rapi i.gender*time, i(id) normal
    
72. 
    
73. ******************
    
74. /* GEE Poisson */
    
75. *****************
    
76. xtset id time
    
77. 
    
78. xi: xtgee rapi i.gender*time, family(poisson) corr(exchangeable)
    
79. 
    
80. ***************************
    
81. /* GEE negative binomial */
    
82. ***************************
    
83. 
    
84. xi: xtgee rapi i.gender*time, family(nbinomial) corr(exchangeable)

复制代码

谢谢分享！

tutorial on count regression and zero-altered count models for longitudinal addictions data.谢谢分享！