Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • gologit parallel*regression assumption violated; brant vs BIC

    Dear Statalists,

    I have a question related to the parallel regression assumption in a ordered logit model.

    My current models are estimated using ologit. But brant test shows the parallel regression assumption has been violated. I was wondering if this is due my sample size (over 7000).

    I considered using a BIC test with gologit2 instead of brant. BIC test prefers ologit over gologit (AIC favours gologit).

    However, a warning of having "an outcome with a predicted probability that is less than 0" showed up. The number of cases can be as high as 10% of the total sample size.

    Questions:
    1. Is it justifiable to use the BIC test and ignore the warning? gologit, npl is not used to estimate coefficients or p-values. My plan is only using it as BIC test.

    2. I prefer using ologit. By any chance can I ignore the brant test and keep ologit ?
    I will report the brant test in footnotes. My outcome variables are 5-level social class and educational qualifications.

    ologit
    Code:
    gologit2 va i.vb i.v1 i.v2 i.v3 i.v4 i.v5 c.v6 ///
    >         c.v7 c.v7##c.v7 i.v8 i.v9 i.v10, ///
    >         pl   sto(ologit) nolog
    note: v7 omitted because of collinearity
    
    Generalized Ordered Logit Estimates             Number of obs     =      7,279
                                                    LR chi2(28)       =    1941.19
                                                    Prob > chi2       =     0.0000
    Log likelihood =  -9545.002                     Pseudo R2         =     0.0923
    
     ( 1)  [1]2.vb - [2]2.vb = 0
     ( 2)  [1]3.vb - [2]3.vb = 0
     ( 3)  [1]4.vb - [2]4.vb = 0
     ( 4)  [1]5.vb - [2]5.vb = 0
     ( 5)  [1]2.v1 - [2]2.v1 = 0
     ( 6)  [1]3.v1 - [2]3.v1 = 0
     ( 7)  [1]4.v1 - [2]4.v1 = 0
     ( 8)  [1]5.v1 - [2]5.v1 = 0
     ( 9)  [1]2.v2 - [2]2.v2 = 0
     (10)  [1]3.v2 - [2]3.v2 = 0
     (11)  [1]4.v2 - [2]4.v2 = 0
     (12)  [1]5.v2 - [2]5.v2 = 0
     (13)  [1]2.v3 - [2]2.v3 = 0
     (14)  [1]3.v3 - [2]3.v3 = 0
     (15)  [1]4.v3 - [2]4.v3 = 0
     (16)  [1]5.v3 - [2]5.v3 = 0
     (17)  [1]2.v4 - [2]2.v4 = 0
     (18)  [1]3.v4 - [2]3.v4 = 0
     (19)  [1]4.v4 - [2]4.v4 = 0
     (20)  [1]5.v4 - [2]5.v4 = 0
     (21)  [1]1.v5 - [2]1.v5 = 0
     (22)  [1]v6 - [2]v6 = 0
     (23)  [1]v7 - [2]v7 = 0
     (24)  [1]c.v7#c.v7 - [2]c.v7#c.v7 = 0
     (25)  [1]2.v8 - [2]2.v8 = 0
     (26)  [1]3.v8 - [2]3.v8 = 0
     (27)  [1]2.v9 - [2]2.v9 = 0
     (28)  [1]2.v10 - [2]2.v10 = 0
     (29)  [2]2.vb - [3]2.vb = 0
     (30)  [2]3.vb - [3]3.vb = 0
     (31)  [2]4.vb - [3]4.vb = 0
     (32)  [2]5.vb - [3]5.vb = 0
     (33)  [2]2.v1 - [3]2.v1 = 0
     (34)  [2]3.v1 - [3]3.v1 = 0
     (35)  [2]4.v1 - [3]4.v1 = 0
     (36)  [2]5.v1 - [3]5.v1 = 0
     (37)  [2]2.v2 - [3]2.v2 = 0
     (38)  [2]3.v2 - [3]3.v2 = 0
     (39)  [2]4.v2 - [3]4.v2 = 0
     (40)  [2]5.v2 - [3]5.v2 = 0
     (41)  [2]2.v3 - [3]2.v3 = 0
     (42)  [2]3.v3 - [3]3.v3 = 0
     (43)  [2]4.v3 - [3]4.v3 = 0
     (44)  [2]5.v3 - [3]5.v3 = 0
     (45)  [2]2.v4 - [3]2.v4 = 0
     (46)  [2]3.v4 - [3]3.v4 = 0
     (47)  [2]4.v4 - [3]4.v4 = 0
     (48)  [2]5.v4 - [3]5.v4 = 0
     (49)  [2]1.v5 - [3]1.v5 = 0
     (50)  [2]v6 - [3]v6 = 0
     (51)  [2]v7 - [3]v7 = 0
     (52)  [2]c.v7#c.v7 - [3]c.v7#c.v7 = 0
     (53)  [2]2.v8 - [3]2.v8 = 0
     (54)  [2]3.v8 - [3]3.v8 = 0
     (55)  [2]2.v9 - [3]2.v9 = 0
     (56)  [2]2.v10 - [3]2.v10 = 0
     (57)  [3]2.vb - [4]2.vb = 0
     (58)  [3]3.vb - [4]3.vb = 0
     (59)  [3]4.vb - [4]4.vb = 0
     (60)  [3]5.vb - [4]5.vb = 0
     (61)  [3]2.v1 - [4]2.v1 = 0
     (62)  [3]3.v1 - [4]3.v1 = 0
     (63)  [3]4.v1 - [4]4.v1 = 0
     (64)  [3]5.v1 - [4]5.v1 = 0
     (65)  [3]2.v2 - [4]2.v2 = 0
     (66)  [3]3.v2 - [4]3.v2 = 0
     (67)  [3]4.v2 - [4]4.v2 = 0
     (68)  [3]5.v2 - [4]5.v2 = 0
     (69)  [3]2.v3 - [4]2.v3 = 0
     (70)  [3]3.v3 - [4]3.v3 = 0
     (71)  [3]4.v3 - [4]4.v3 = 0
     (72)  [3]5.v3 - [4]5.v3 = 0
     (73)  [3]2.v4 - [4]2.v4 = 0
     (74)  [3]3.v4 - [4]3.v4 = 0
     (75)  [3]4.v4 - [4]4.v4 = 0
     (76)  [3]5.v4 - [4]5.v4 = 0
     (77)  [3]1.v5 - [4]1.v5 = 0
     (78)  [3]v6 - [4]v6 = 0
     (79)  [3]v7 - [4]v7 = 0
     (80)  [3]c.v7#c.v7 - [4]c.v7#c.v7 = 0
     (81)  [3]2.v8 - [4]2.v8 = 0
     (82)  [3]3.v8 - [4]3.v8 = 0
     (83)  [3]2.v9 - [4]2.v9 = 0
     (84)  [3]2.v10 - [4]2.v10 = 0
    ------------------------------------------------------------------------------
              va |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
    -------------+----------------------------------------------------------------
    1            |
              vb |
              2  |   .9354747   .0916725    10.20   0.000        .7558    1.115149
              3  |   1.330913   .0643736    20.67   0.000     1.204743    1.457083
              4  |   1.824894   .0692276    26.36   0.000     1.689211    1.960578
              5  |   2.567934   .0920257    27.90   0.000     2.387567    2.748301
                 |
              v1 |
              2  |   .0510374   .0898897     0.57   0.570    -.1251431    .2272179
              3  |   .1460976   .0763766     1.91   0.056    -.0035979     .295793
              4  |   .0881399   .0765471     1.15   0.250    -.0618896    .2381694
              5  |   .1695604   .0663425     2.56   0.011     .0395315    .2995894
                 |
              v2 |
              2  |   .1110186   .0695059     1.60   0.110    -.0252106    .2472477
              3  |   .0503031   .0765351     0.66   0.511    -.0997029    .2003091
              4  |   .2182135   .0814266     2.68   0.007     .0586203    .3778066
              5  |   .3616781   .0719075     5.03   0.000     .2207421    .5026142
                 |
              v3 |
              2  |   .0735787   .1019591     0.72   0.471    -.1262574    .2734149
              3  |   .1311861   .0866132     1.51   0.130    -.0385727    .3009449
              4  |   .1515616   .0856373     1.77   0.077    -.0162843    .3194076
              5  |   .2269051   .0841826     2.70   0.007     .0619102       .3919
                 |
              v4 |
              2  |    .008858   .1030983     0.09   0.932    -.1932109    .2109269
              3  |  -.1552937   .0972472    -1.60   0.110    -.3458946    .0353073
              4  |  -.2228292   .0894085    -2.49   0.013    -.3980666   -.0475918
              5  |  -.0138213    .089547    -0.15   0.877    -.1893302    .1616876
                 |
              v5 |
              1  |  -.2589323   .0618058    -4.19   0.000    -.3800695    -.137795
                 |
              v6 |  -.2440731     .04129    -5.91   0.000    -.3250001   -.1631462
              v7 |  -.1243455   .0249541    -4.98   0.000    -.1732546   -.0754365
              v7 |   2.13e-16  (omitted)
                 |
       c.v7#c.v7 |   .0014062   .0003278     4.29   0.000     .0007637    .0020487
                 |
              v8 |
              2  |  -.1233608   .0884532    -1.39   0.163    -.2967259    .0500043
              3  |  -.2117187   .1053018    -2.01   0.044    -.4181064    -.005331
                 |
              v9 |
              2  |   .5615378   .0564712     9.94   0.000     .4508563    .6722194
                 |
             v10 |
              2  |  -.1832669   .0455544    -4.02   0.000     -.272552   -.0939819
                 |
           _cons |   3.725493   .5789558     6.43   0.000     2.590761    4.860226
    -------------+----------------------------------------------------------------
    2            |
              vb |
              2  |   .9354747   .0916725    10.20   0.000        .7558    1.115149
              3  |   1.330913   .0643736    20.67   0.000     1.204743    1.457083
              4  |   1.824894   .0692276    26.36   0.000     1.689211    1.960578
              5  |   2.567934   .0920257    27.90   0.000     2.387567    2.748301
                 |
              v1 |
              2  |   .0510374   .0898897     0.57   0.570    -.1251431    .2272179
              3  |   .1460976   .0763766     1.91   0.056    -.0035979     .295793
              4  |   .0881399   .0765471     1.15   0.250    -.0618896    .2381694
              5  |   .1695604   .0663425     2.56   0.011     .0395315    .2995894
                 |
              v2 |
              2  |   .1110186   .0695059     1.60   0.110    -.0252106    .2472477
              3  |   .0503031   .0765351     0.66   0.511    -.0997029    .2003091
              4  |   .2182135   .0814266     2.68   0.007     .0586203    .3778066
              5  |   .3616781   .0719075     5.03   0.000     .2207421    .5026142
                 |
              v3 |
              2  |   .0735787   .1019591     0.72   0.471    -.1262574    .2734149
              3  |   .1311861   .0866132     1.51   0.130    -.0385727    .3009449
              4  |   .1515616   .0856373     1.77   0.077    -.0162843    .3194076
              5  |   .2269051   .0841826     2.70   0.007     .0619102       .3919
                 |
              v4 |
              2  |    .008858   .1030983     0.09   0.932    -.1932109    .2109269
              3  |  -.1552937   .0972472    -1.60   0.110    -.3458946    .0353073
              4  |  -.2228292   .0894085    -2.49   0.013    -.3980666   -.0475918
              5  |  -.0138213    .089547    -0.15   0.877    -.1893302    .1616876
                 |
              v5 |
              1  |  -.2589323   .0618058    -4.19   0.000    -.3800695    -.137795
                 |
              v6 |  -.2440731     .04129    -5.91   0.000    -.3250001   -.1631462
              v7 |  -.1243455   .0249541    -4.98   0.000    -.1732546   -.0754365
              v7 |   2.50e-17  (omitted)
                 |
       c.v7#c.v7 |   .0014062   .0003278     4.29   0.000     .0007637    .0020487
                 |
              v8 |
              2  |  -.1233608   .0884532    -1.39   0.163    -.2967259    .0500043
              3  |  -.2117187   .1053018    -2.01   0.044    -.4181064    -.005331
                 |
              v9 |
              2  |   .5615378   .0564712     9.94   0.000     .4508563    .6722194
                 |
             v10 |
              2  |  -.1832669   .0455544    -4.02   0.000     -.272552   -.0939819
                 |
           _cons |   2.861834   .5784971     4.95   0.000     1.728001    3.995667
    -------------+----------------------------------------------------------------
    3            |
              vb |
              2  |   .9354747   .0916725    10.20   0.000        .7558    1.115149
              3  |   1.330913   .0643736    20.67   0.000     1.204743    1.457083
              4  |   1.824894   .0692276    26.36   0.000     1.689211    1.960578
              5  |   2.567934   .0920257    27.90   0.000     2.387567    2.748301
                 |
              v1 |
              2  |   .0510374   .0898897     0.57   0.570    -.1251431    .2272179
              3  |   .1460976   .0763766     1.91   0.056    -.0035979     .295793
              4  |   .0881399   .0765471     1.15   0.250    -.0618896    .2381694
              5  |   .1695604   .0663425     2.56   0.011     .0395315    .2995894
                 |
              v2 |
              2  |   .1110186   .0695059     1.60   0.110    -.0252106    .2472477
              3  |   .0503031   .0765351     0.66   0.511    -.0997029    .2003091
              4  |   .2182135   .0814266     2.68   0.007     .0586203    .3778066
              5  |   .3616781   .0719075     5.03   0.000     .2207421    .5026142
                 |
              v3 |
              2  |   .0735787   .1019591     0.72   0.471    -.1262574    .2734149
              3  |   .1311861   .0866132     1.51   0.130    -.0385727    .3009449
              4  |   .1515616   .0856373     1.77   0.077    -.0162843    .3194076
              5  |   .2269051   .0841826     2.70   0.007     .0619102       .3919
                 |
              v4 |
              2  |    .008858   .1030983     0.09   0.932    -.1932109    .2109269
              3  |  -.1552937   .0972472    -1.60   0.110    -.3458946    .0353073
              4  |  -.2228292   .0894085    -2.49   0.013    -.3980666   -.0475918
              5  |  -.0138213    .089547    -0.15   0.877    -.1893302    .1616876
                 |
              v5 |
              1  |  -.2589323   .0618058    -4.19   0.000    -.3800695    -.137795
                 |
              v6 |  -.2440731     .04129    -5.91   0.000    -.3250001   -.1631462
              v7 |  -.1243455   .0249541    -4.98   0.000    -.1732546   -.0754365
              v7 |  -1.64e-16  (omitted)
                 |
       c.v7#c.v7 |   .0014062   .0003278     4.29   0.000     .0007637    .0020487
                 |
              v8 |
              2  |  -.1233608   .0884532    -1.39   0.163    -.2967259    .0500043
              3  |  -.2117187   .1053018    -2.01   0.044    -.4181064    -.005331
                 |
              v9 |
              2  |   .5615378   .0564712     9.94   0.000     .4508563    .6722194
                 |
             v10 |
              2  |  -.1832669   .0455544    -4.02   0.000     -.272552   -.0939819
                 |
           _cons |   2.425186   .5782236     4.19   0.000     1.291889    3.558484
    -------------+----------------------------------------------------------------
    4            |
              vb |
              2  |   .9354747   .0916725    10.20   0.000        .7558    1.115149
              3  |   1.330913   .0643736    20.67   0.000     1.204743    1.457083
              4  |   1.824894   .0692276    26.36   0.000     1.689211    1.960578
              5  |   2.567934   .0920257    27.90   0.000     2.387567    2.748301
                 |
              v1 |
              2  |   .0510374   .0898897     0.57   0.570    -.1251431    .2272179
              3  |   .1460976   .0763766     1.91   0.056    -.0035979     .295793
              4  |   .0881399   .0765471     1.15   0.250    -.0618896    .2381694
              5  |   .1695604   .0663425     2.56   0.011     .0395315    .2995894
                 |
              v2 |
              2  |   .1110186   .0695059     1.60   0.110    -.0252106    .2472477
              3  |   .0503031   .0765351     0.66   0.511    -.0997029    .2003091
              4  |   .2182135   .0814266     2.68   0.007     .0586203    .3778066
              5  |   .3616781   .0719075     5.03   0.000     .2207421    .5026142
                 |
              v3 |
              2  |   .0735787   .1019591     0.72   0.471    -.1262574    .2734149
              3  |   .1311861   .0866132     1.51   0.130    -.0385727    .3009449
              4  |   .1515616   .0856373     1.77   0.077    -.0162843    .3194076
              5  |   .2269051   .0841826     2.70   0.007     .0619102       .3919
                 |
              v4 |
              2  |    .008858   .1030983     0.09   0.932    -.1932109    .2109269
              3  |  -.1552937   .0972472    -1.60   0.110    -.3458946    .0353073
              4  |  -.2228292   .0894085    -2.49   0.013    -.3980666   -.0475918
              5  |  -.0138213    .089547    -0.15   0.877    -.1893302    .1616876
                 |
              v5 |
              1  |  -.2589323   .0618058    -4.19   0.000    -.3800695    -.137795
                 |
              v6 |  -.2440731     .04129    -5.91   0.000    -.3250001   -.1631462
              v7 |  -.1243455   .0249541    -4.98   0.000    -.1732546   -.0754365
              v7 |  -1.32e-40  (omitted)
                 |
       c.v7#c.v7 |   .0014062   .0003278     4.29   0.000     .0007637    .0020487
                 |
              v8 |
              2  |  -.1233608   .0884532    -1.39   0.163    -.2967259    .0500043
              3  |  -.2117187   .1053018    -2.01   0.044    -.4181064    -.005331
                 |
              v9 |
              2  |   .5615378   .0564712     9.94   0.000     .4508563    .6722194
                 |
             v10 |
              2  |  -.1832669   .0455544    -4.02   0.000     -.272552   -.0939819
                 |
           _cons |   1.939493   .5779007     3.36   0.001     .8068289    3.072158
    ------------------------------------------------------------------------------
    
    .
    gologit
    Code:
    . gologit2 va i.vb i.v1 i.v2 i.v3 i.v4 i.v5 c.v6 ///
    >         c.v7 c.v7##c.v7 i.v8 i.v9 i.v10, ///
    >         npl   sto(gologit)
    note: v7 omitted because of collinearity
    
    Generalized Ordered Logit Estimates             Number of obs     =      7,279
                                                    LR chi2(112)      =    2538.64
                                                    Prob > chi2       =     0.0000
    Log likelihood =  -9246.277                     Pseudo R2         =     0.1207
    
    ------------------------------------------------------------------------------
              va |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
    -------------+----------------------------------------------------------------
    1            |
              vb |
              2  |   1.057546   .1015163    10.42   0.000     .8585773    1.256514
              3  |   1.378632   .0700381    19.68   0.000      1.24136    1.515904
              4  |   1.968494   .0802388    24.53   0.000     1.811229    2.125759
              5  |   2.412775   .1206646    20.00   0.000     2.176277    2.649274
                 |
              v1 |
              2  |   .0011858   .1009543     0.01   0.991    -.1966811    .1990527
              3  |   .1286831   .0883769     1.46   0.145    -.0445323    .3018986
              4  |   .0889813   .0870517     1.02   0.307    -.0816369    .2595996
              5  |   .1769239   .0766116     2.31   0.021     .0267679    .3270799
                 |
              v2 |
              2  |   .0981972   .0810137     1.21   0.225    -.0605867    .2569811
              3  |   .0451554   .0930535     0.49   0.627     -.137226    .2275368
              4  |   .1892947   .1000678     1.89   0.059    -.0068346     .385424
              5  |   .3006894   .0889552     3.38   0.001     .1263404    .4750384
                 |
              v3 |
              2  |  -.0499873   .1120794    -0.45   0.656    -.2696588    .1696843
              3  |   .0058835   .0969007     0.06   0.952    -.1840385    .1958054
              4  |    .108186   .0972289     1.11   0.266    -.0823792    .2987512
              5  |   .1546271   .0960621     1.61   0.107    -.0336511    .3429052
                 |
              v4 |
              2  |  -.0454071   .1138384    -0.40   0.690    -.2685264    .1777121
              3  |  -.1274106   .1103419    -1.15   0.248    -.3436768    .0888555
              4  |  -.1917055   .1016416    -1.89   0.059    -.3909193    .0075083
              5  |   .0114948   .1032079     0.11   0.911     -.190789    .2137785
                 |
              v5 |
              1  |  -.2770506   .0788353    -3.51   0.000     -.431565   -.1225362
                 |
              v6 |  -.2819923   .0505387    -5.58   0.000    -.3810464   -.1829382
              v7 |  -.1596512   .0344204    -4.64   0.000     -.227114   -.0921885
              v7 |          0  (omitted)
                 |
       c.v7#c.v7 |   .0018167   .0004687     3.88   0.000     .0008982    .0027353
                 |
              v8 |
              2  |  -.0222869   .1076871    -0.21   0.836    -.2333496    .1887759
              3  |  -.1956593   .1261278    -1.55   0.121    -.4428652    .0515466
                 |
              v9 |
              2  |   .6745173   .0679417     9.93   0.000      .541354    .8076806
                 |
             v10 |
              2  |  -.0312754   .0536759    -0.58   0.560    -.1364783    .0739276
                 |
           _cons |   4.599815   .7444705     6.18   0.000      3.14068    6.058951
    -------------+----------------------------------------------------------------
    2            |
              vb |
              2  |   .8702234   .1082148     8.04   0.000     .6581263     1.08232
              3  |   1.395972   .0758671    18.40   0.000     1.247275    1.544668
              4  |   1.926601   .0812303    23.72   0.000     1.767393     2.08581
              5  |   2.700606   .1113845    24.25   0.000     2.482297    2.918916
                 |
              v1 |
              2  |   .0638583   .1024715     0.62   0.533    -.1369821    .2646987
              3  |   .1779957   .0883522     2.01   0.044     .0048285    .3511629
              4  |   .0808547    .087243     0.93   0.354    -.0901384    .2518478
              5  |   .1562247   .0761626     2.05   0.040     .0069487    .3055007
                 |
              v2 |
              2  |   .1563776   .0800952     1.95   0.051    -.0006061    .3133614
              3  |   .2080237    .089178     2.33   0.020      .033238    .3828094
              4  |   .1410086   .0946112     1.49   0.136     -.044426    .3264432
              5  |   .3652043   .0830363     4.40   0.000     .2024562    .5279525
                 |
              v3 |
              2  |   .1730991   .1187941     1.46   0.145    -.0597331    .4059313
              3  |   .2305115    .101549     2.27   0.023     .0314791    .4295439
              4  |   .1637572   .1015522     1.61   0.107    -.0352815     .362796
              5  |   .3794032   .0994852     3.81   0.000     .1844157    .5743906
                 |
              v4 |
              2  |   .1365176   .1194626     1.14   0.253    -.0976247      .37066
              3  |  -.1373012   .1138338    -1.21   0.228    -.3604113    .0858089
              4  |  -.2655482   .1047239    -2.54   0.011    -.4708032   -.0602931
              5  |  -.0455416   .1047401    -0.43   0.664    -.2508284    .1597452
                 |
              v5 |
              1  |  -.1893975   .0707849    -2.68   0.007    -.3281333   -.0506616
                 |
              v6 |  -.2345988   .0482494    -4.86   0.000    -.3291659   -.1400317
              v7 |  -.0910479   .0305284    -2.98   0.003    -.1508825   -.0312132
              v7 |          0  (omitted)
                 |
       c.v7#c.v7 |   .0009717   .0004086     2.38   0.017     .0001709    .0017725
                 |
              v8 |
              2  |  -.2180263   .1011086    -2.16   0.031    -.4161955   -.0198571
              3  |  -.3321529   .1197928    -2.77   0.006    -.5669424   -.0973634
                 |
              v9 |
              2  |   .5519539   .0654376     8.43   0.000     .4236985    .6802092
                 |
             v10 |
              2  |  -.5764329   .0528799   -10.90   0.000    -.6800756   -.4727902
                 |
           _cons |   2.262614   .6861766     3.30   0.001      .917733    3.607496
    -------------+----------------------------------------------------------------
    3            |
              vb |
              2  |   .8592302   .1159131     7.41   0.000     .6320446    1.086416
              3  |   1.303388   .0835124    15.61   0.000     1.139707    1.467069
              4  |   1.741451     .08666    20.10   0.000       1.5716    1.911301
              5  |   2.626693   .1095208    23.98   0.000     2.412036    2.841349
                 |
              v1 |
              2  |   .1050732   .1068755     0.98   0.326    -.1043989    .3145452
              3  |   .0994135   .0927711     1.07   0.284    -.0824146    .2812415
              4  |    .113478   .0903037     1.26   0.209     -.063514    .2904699
              5  |   .2054129   .0789167     2.60   0.009      .050739    .3600868
                 |
              v2 |
              2  |   .1466935   .0839244     1.75   0.080    -.0177954    .3111824
              3  |  -.0280507   .0921554    -0.30   0.761     -.208672    .1525707
              4  |   .4429523    .096042     4.61   0.000     .2547135    .6311911
              5  |   .4687237   .0840789     5.57   0.000     .3039321    .6335153
                 |
              v3 |
              2  |   .0557613   .1250663     0.45   0.656    -.1893641    .3008867
              3  |   .0953552   .1078084     0.88   0.376    -.1159453    .3066557
              4  |    .058259   .1076718     0.54   0.588    -.1527738    .2692918
              5  |   .1263777   .1059007     1.19   0.233    -.0811839    .3339393
                 |
              v4 |
              2  |   .0740139   .1263452     0.59   0.558    -.1736182    .3216459
              3  |  -.2135359   .1200103    -1.78   0.075    -.4487518    .0216801
              4  |  -.2460635   .1098843    -2.24   0.025    -.4614327   -.0306943
              5  |  -.0571257   .1091305    -0.52   0.601    -.2710175    .1567661
                 |
              v5 |
              1  |  -.2823429   .0701502    -4.02   0.000    -.4198348   -.1448509
                 |
              v6 |  -.2416987   .0491156    -4.92   0.000    -.3379635   -.1454339
              v7 |  -.1229942   .0289019    -4.26   0.000    -.1796409   -.0663474
              v7 |          0  (omitted)
                 |
       c.v7#c.v7 |   .0013681   .0003788     3.61   0.000     .0006257    .0021105
                 |
              v8 |
              2  |  -.1567548   .1011132    -1.55   0.121     -.354933    .0414234
              3  |  -.2620699   .1197713    -2.19   0.029    -.4968174   -.0273223
                 |
              v9 |
              2  |   .5755259   .0673112     8.55   0.000     .4435984    .7074535
                 |
             v10 |
              2  |  -.3686982   .0548224    -6.73   0.000    -.4761482   -.2612482
                 |
           _cons |   2.624467    .676144     3.88   0.000     1.299249    3.949684
    -------------+----------------------------------------------------------------
    4            |
              vb |
              2  |   .4850696     .13019     3.73   0.000      .229902    .7402373
              3  |   .8898708   .0924665     9.62   0.000     .7086399    1.071102
              4  |   1.353353   .0943561    14.34   0.000     1.168418    1.538288
              5  |   2.135837   .1117459    19.11   0.000     1.916819    2.354855
                 |
              v1 |
              2  |   .1139656   .1139944     1.00   0.317    -.1094594    .3373905
              3  |   .1805241   .0991742     1.82   0.069    -.0138537    .3749019
              4  |   .1122109   .0963465     1.16   0.244    -.0766249    .3010466
              5  |   .1810273    .084731     2.14   0.033     .0149576    .3470971
                 |
              v2 |
              2  |   .0378617    .090632     0.42   0.676    -.1397738    .2154971
              3  |  -.0448437   .0982323    -0.46   0.648    -.2373754     .147688
              4  |   .2233114   .1013877     2.20   0.028     .0245951    .4220277
              5  |   .4133253   .0881971     4.69   0.000     .2404622    .5861883
                 |
              v3 |
              2  |   .3657221   .1380484     2.65   0.008     .0951522     .636292
              3  |   .4106705   .1184537     3.47   0.001     .1785054    .6428355
              4  |   .3429813    .119856     2.86   0.004     .1080678    .5778948
              5  |   .3502806   .1168685     3.00   0.003     .1212226    .5793386
                 |
              v4 |
              2  |   .0784126   .1376942     0.57   0.569    -.1914631    .3482883
              3  |  -.1930729   .1293732    -1.49   0.136    -.4466397    .0604939
              4  |  -.2242521   .1194766    -1.88   0.061    -.4584219    .0099177
              5  |  -.0004533   .1152395    -0.00   0.997    -.2263185    .2254119
                 |
              v5 |
              1  |  -.2926428   .0719061    -4.07   0.000    -.4335762   -.1517093
                 |
              v6 |   -.212524   .0527156    -4.03   0.000    -.3158446   -.1092033
              v7 |  -.1021918   .0287857    -3.55   0.000    -.1586107   -.0457728
              v7 |          0  (omitted)
                 |
       c.v7#c.v7 |    .001206   .0003707     3.25   0.001     .0004795    .0019326
                 |
              v8 |
              2  |  -.1991453   .1056061    -1.89   0.059    -.4061294    .0078389
              3  |  -.1294105   .1253291    -1.03   0.302    -.3750511    .1162301
                 |
              v9 |
              2  |   .4131756   .0725179     5.70   0.000     .2710432     .555308
                 |
             v10 |
              2  |  -.0839719   .0582363    -1.44   0.149    -.1981129     .030169
                 |
           _cons |   1.445716   .7027124     2.06   0.040     .0684246    2.823007
    ------------------------------------------------------------------------------
    
    WARNING! 803 in-sample cases have an outcome with a predicted probability that is
    less than 0. See the gologit2 help section on Warning Messages for more information.
    
    .
    end of do-file
    LRTEST
    Code:
    . lrtest ologit gologit, stat  
    
    Likelihood-ratio test                                 LR chi2(84) =    597.45
    (Assumption: ologit nested in gologit)                Prob > chi2 =    0.0000
    
    Akaike's information criterion and Bayesian information criterion
    
    -----------------------------------------------------------------------------
           Model |        Obs  ll(null)  ll(model)      df         AIC        BIC
    -------------+---------------------------------------------------------------
          ologit |      7,279  -10515.6  -9545.002      32       19154   19374.57
         gologit |      7,279  -10515.6  -9246.277     116    18724.55   19524.11
    -----------------------------------------------------------------------------
                   Note: N=Obs used in calculating BIC; see [R] BIC note.
    
    .
    Brant test after ologit
    Code:
     brant
    
    Brant test of parallel regression assumption
    
                  |       chi2     p>chi2      df
     -------------+------------------------------
              All |     714.78      0.000      84
     -------------+------------------------------
             2.vb |      27.60      0.000       3
             3.vb |      65.12      0.000       3
             4.vb |      71.25      0.000       3
             5.vb |      41.76      0.000       3
             2.v1 |       1.82      0.612       3
             3.v1 |       9.06      0.029       3
             4.v1 |       2.93      0.403       3
             5.v1 |       0.78      0.855       3
             2.v2 |       0.94      0.815       3
             3.v2 |      23.48      0.000       3
             4.v2 |      20.21      0.000       3
             5.v2 |       5.41      0.144       3
             2.v3 |       9.05      0.029       3
             3.v3 |      22.07      0.000       3
             4.v3 |       6.32      0.097       3
             5.v3 |      28.54      0.000       3
             2.v4 |       7.70      0.053       3
             3.v4 |       8.74      0.033       3
             4.v4 |       6.76      0.080       3
             5.v4 |       5.99      0.112       3
             1.v5 |       5.87      0.118       3
               v6 |       1.03      0.794       3
               v7 |      22.64      0.000       3
        c.v7#c.v7 |      15.66      0.001       3
             2.v8 |       6.71      0.082       3
             3.v8 |       5.26      0.154       3
             2.v9 |      10.55      0.014       3
            2.v10 |     262.59      0.000       3
    
    A significant test statistic provides evidence that the parallel
    regression assumption has been violated.
    Last edited by Min Zhang; 26 Oct 2016, 12:59.

  • #2
    Hi Min. Using a BIC test as the rationale for choosing a more parsimonious model would not be unusual. However, it seems like you have a lot of variables in the model, many of which are insignificant even in the ologit version. Further, if you did give up on ologit, it would be very unlikely that you would switch to a purely unconstrained gologit model. More likely, you would switch to partial proportional odds, where some variables were constrained and some were not. The brant, detail option, or gologit2's autofit option could help to identify variables that are candidates for having the proprtional odds assumption relaxed.

    In short, I think you have a reasonable case for ologit. But personally, I would fiddle around with the model a bit more and see if I really want all the vars in it that I have now; and if I do want them would it help if I selectively relaxed the PO assumption for some vars but not all of them.
    -------------------------------------------
    Richard Williams, Notre Dame Dept of Sociology
    StataNow Version: 19.5 MP (2 processor)

    EMAIL: [email protected]
    WWW: https://academicweb.nd.edu/~rwilliam/

    Comment


    • #3
      Hi Richard,
      Many thanks for your reply.

      Yes, I would like to keep these variables. One of the goals of the research is to show that some variables are statistically significant but some are not, once these variables are taken into account together.

      I thought about your suggestions and tried gologit2, autofit.

      1. The results of gologit, autofit shows that variables of interest (which are important to my research questions) passed the Wald test and have proportionality constraints imposed. Those variables which failed to pass the test (i.e., violated the parallel assumption) are control variables which fall off the radar of my research questions.

      Together with the BIC test I mentioned earlier, do you think whether they can justify using ologit? Of course I will make it clear in appendix or footnotes.

      2. With gologit2, autofit, one of my model still reports a worrying number of cases which have an outcome with a predicted probability less than 0.
      I read your posts regarding this issues but I do not want to further collapse the outcome variable (Ideally the outcome variable is 7-level; now I collapsed it into 5-level).
      Once again, I was wondering if the BIC test is still valid while many cases have negative predicted probabilities.


      Many thanks again.

      BTW, I work with Stata 14.

      Comment


      • #4
        Another question, I wonder why BIC and AIC give different results here?

        I understand this is actually pretty common and have a rough idea that BIC and AIC suggest different things in terms of model fit.

        But what I don't understand is why they favour different models here.

        Comment


        • #5
          A large number of negative predicted probabilities would make me worry about gologit. That doesn't mean I would therefore stick with ologit but it might make me consider alternative approaches, e.g. mlogit.

          I can't see your descriptive statistics, but I might worry that you are spreading the data too thin. Maybe some categories with small Ns should be combined with others.

          It isn't that unusual for BIC and AIC to lead to different conclusions. I wouldn't worry too much about it. My bigger concern would be whether I have the model correctly specified in the first place.

          I suspect nothing too horrible will happen if you stick with ologit. But you can always try things a couple of different ways and assess whether it makes much difference in your conclusions.
          -------------------------------------------
          Richard Williams, Notre Dame Dept of Sociology
          StataNow Version: 19.5 MP (2 processor)

          EMAIL: [email protected]
          WWW: https://academicweb.nd.edu/~rwilliam/

          Comment


          • #6
            Hi Richard,

            Thank you for your reply.

            First, I do have small sample sizes in some categories. The outcome variable here is social class. Ideally I would like to use 7-level class classification but due to practical consideration, I use 5-level social class now. I tried the model on 3-level social class and it worked well. But I wish I do not have to do so. Using 3-level social class as outcome variable would lose too much information.

            After the gologit, npl,

            Code:
            gen sam1=e(sample)
            
            . predict pr1 pr2 pr3 pr4 pr5 if sam==1
            (option p assumed; predicted probabilities)
            (4664 missing values generated)
            
            . su pr?
            
                Variable |        Obs        Mean    Std. Dev.       Min        Max
            -------------+---------------------------------------------------------
                     pr1 |      7,279    .3849712    .2152725   .0088281   .8681916
                     pr2 |      7,279    .1681077    .0814216  -.0200286   .4031219
                     pr3 |      7,279    .0844578    .0550674  -.0690993    .281536
                     pr4 |      7,279    .0878238    .0634541  -.0343109   .2848485
                     pr5 |      7,279    .2746395    .1643656    .035751   .8774264
            
            . tab va if sam==1
            
                      va |      Freq.     Percent        Cum.
            ------------+-----------------------------------
                      1 |      2,793       38.37       38.37
                      2 |      1,229       16.88       55.25
                      3 |        610        8.38       63.64
                      4 |        640        8.79       72.43
                      5 |      2,007       27.57      100.00
            ------------+-----------------------------------
                  Total |      7,279      100.00
            2. I have also tried mlogit before. As expected, it did not make any essential differences.

            I am worried if I should proceed the ologit analyses. Excuse my lack of knowledge, I failed to find any superior solution or models in this case.
            Or is there anything I can do to make using ologit more legitimate?


            Thanks a lot,
            Min
            Last edited by Min Zhang; 27 Oct 2016, 13:43.

            Comment

            Working...
            X