Announcement

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

  • Insignificant constant and Chi2 in Multinomial regression

    Hello,

    I am trying to use Multinomail probit model for a binary variable (had treatment or not). Unluckily the estimated parameters are not significant and the constant and the Chi2 are not significant too.
    Kindly the results are displayed below, Please do have a look and your suggestion is much welcomed.

    One more question, I am using Stata version 15. I am conducting Heckprobit model where I want to calculate miles ratio to use it as independent variable in other regression model. Any help with this would be highly appreciated.

    My syntax:
    --------------
    mprobit no_treatment s614a female two_child many_child agegrp2 agegrp3 agegrp4 agegrp5 agegrp6 agegrp7 priedu secedu hgedu fpriedu fsecedu fhgedu wipoorer wimiddle wiricher wirichest earning v714 fprofessionls fagr_service fmanual insured v481a v481b v481c urban

    -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

    Code and Results:

    [. mprobit no_treatment s614a female two_child many_child ///
    > agegrp2 agegrp3 agegrp4 agegrp5 agegrp6 agegrp7 ///
    > priedu secedu hgedu fpriedu fsecedu fhgedu ///
    > wipoorer wimiddle wiricher wirichest earning v714 ///
    > fprofessionls fagr_service fmanual ///
    > insured v481a v481b v481c urban

    Iteration 0: log likelihood = -132.71612
    Iteration 1: log likelihood = -132.46317
    Iteration 2: log likelihood = -132.46279
    Iteration 3: log likelihood = -132.46279

    Multinomial probit regression Number of obs = 249
    Wald chi2(30) = 37.99
    Log likelihood = -132.46279
    Prob > chi2 = 0.1499
    Last edited by Embarika Farouk; 11 Apr 2018, 09:57.

  • #2
    Embarika, it's easier to read code and results if you present them in the code delimiters. If you use the # button in the formatting toolbar (it's between the buttons labeled " and <>), that will give you a pair of code delimiters. Type your code and results between them.

    You are using a multinomial model for categorical data on what looks like a binary variable. The results should be equivalent, but why not just run a probit or logit regression to begin with? Also, it's usually easier to comprehend the results if 1 represents treatment, and 0 represents not treated.

    That said, in the frequentist paradigm, your results are not statistically significant. It doesn't seem like they would be significant in the Bayesian paradigm, either. Either the treatment didn't do much, or the data were poorly collected, or both.

    If you can comment more specifically about your intended use for -heckprobit-, others will be able to give you help. What specific problems do you have?
    Be aware that it can be very hard to answer a question without sample data. You can use the dataex command for this. Type help dataex at the command line.

    When presenting code or results, please use the code delimiters format them. Use the # button on the formatting toolbar, between the " (double quote) and <> buttons.

    Comment


    • #3
      Hi Weiwen,

      Thank you very much for your kind response.
      Please, see the code below;

      I was advised to use Multinomial Probit. But I am trying to use Probit too.

      My model is to investigate the determinants of child treatment from Diarrhea disease, I have dataset contains 15,448 obs.
      only 2,010 of children had diarrhea and 13,456 did not,

      I am trying to conduct Heckprobit to avoid any selection bias. In another word, to include the children who did not have diarrhea in my analysis as ignoring them will result in selection bias;

      Then I am trying to calculate the 'Miles Ratio' of that model; Heckprobit selection; and add it in my Probit model as explanatory variable.

      in my Probit model; my main variable of interest is 'No treatment' with 1,331 obs. had treatment; and 679 obs. with no treatment


      Code:
       .  mprobit no_treatment s614a female two_child many_child ///
      >  agegrp2 agegrp3 agegrp4 agegrp5 agegrp6 agegrp7 ///
      >  priedu secedu hgedu fpriedu fsecedu fhgedu ///
      >  wipoorer wimiddle wiricher wirichest earning v714 ///
      >  fprofessionls fagr_service fmanual ///
      >  insured v481a v481b v481c urban
      
      Iteration 0:   log likelihood = -132.71612
      Iteration 1:   log likelihood = -132.46317
      Iteration 2:   log likelihood = -132.46279
      Iteration 3:   log likelihood = -132.46279
      
      Multinomial probit regression                   Number of obs     =        249
                                                      Wald chi2(30)     =      37.99
      Log likelihood = -132.46279                     Prob > chi2       =     0.1499
      
      -------------------------------------------------------------------------------
       no_treatment |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
      --------------+----------------------------------------------------------------
      0             |  (base outcome) had treatment
      --------------+----------------------------------------------------------------
      1             |  No treatment
      
              s614a |  -.1085908   .1074892    -1.01   0.312    -.3192657    .1020841
             female |   .5148575   .2736887     1.88   0.060    -.0215626    1.051278
          two_child |   .0307547   .3053929     0.10   0.920    -.5678044    .6293138
         many_child |  -.0474165   .4852137    -0.10   0.922    -.9984178    .9035848
            agegrp2 |  -2.611431    .911632    -2.86   0.004    -4.398197    -.824665
            agegrp3 |   -2.17647   .7331017    -2.97   0.003    -3.613323   -.7396176
            agegrp4 |  -1.881747   .7076972    -2.66   0.008    -3.268808    -.494686
            agegrp5 |  -1.450206   .7224937    -2.01   0.045    -2.866268   -.0341445
            agegrp6 |  -1.091669   .7448166    -1.47   0.143    -2.551483    .3681447
            agegrp7 |  -1.016377   .8689269    -1.17   0.242    -2.719443    .6866883
             priedu |   .4583253   .6231669     0.74   0.462    -.7630593     1.67971
             secedu |   .1952931   .4418257     0.44   0.658    -.6706693    1.061256
              hgedu |   .4687681   .5827397     0.80   0.421    -.6733807    1.610917
            fpriedu |    1.75843    .613171     2.87   0.004     .5566374    2.960224
            fsecedu |   1.338056   .5205958     2.57   0.010     .3177074    2.358405
             fhgedu |   .7009966   .6542251     1.07   0.284    -.5812611    1.983254
           wipoorer |  -.2296063   .4999749    -0.46   0.646    -1.209539    .7503265
           wimiddle |  -.6698629   .4564143    -1.47   0.142    -1.564418    .2246927
           wiricher |   .6891499   .5194902     1.33   0.185    -.3290323    1.707332
          wirichest |     .71413   .6785258     1.05   0.293    -.6157561    2.044016
            earning |  -.2095367   .4021047    -0.52   0.602    -.9976476    .5785741
               v714 |  -.5813936   .7808939    -0.74   0.457    -2.111918    .9491304
      fprofessionls |  -1.508128   1.193424    -1.26   0.206    -3.847196    .8309409
       fagr_service |  -1.193323   1.221807    -0.98   0.329    -3.588021    1.201375
            fmanual |  -1.487676   1.189098    -1.25   0.211    -3.818266    .8429139
            insured |  -.4877266   1.577769    -0.31   0.757    -3.580097    2.604644
              v481a |   .9811404   1.541944     0.64   0.525    -2.041014    4.003295
              v481b |   .3413193   1.509342     0.23   0.821    -2.616937    3.299575
              v481c |   .9630612   1.527076     0.63   0.528    -2.029952    3.956074
              urban |  -.9082531   .4878754    -1.86   0.063    -1.864471    .0479651
              _cons |   1.891489   1.651832     1.15   0.252    -1.346043     5.12902
      -------------------------------------------------------------------------------
      
      .
      end of do-file
      
      .
      Last edited by Embarika Farouk; 11 Apr 2018, 10:49.

      Comment


      • #4
        I notice you seem to have a number of interaction terms in your model.

        It appears that you have included the interaction of gender (female) with employment (professionals, agr_service, manual) without having included employment without the interactions. This is generally not a good idea.

        I agree with post #2 that the use of mprobit seems unnecessary given that you have only two outcomes.

        I would add that your model would be better expressed using factor variable notation rather than creating dummy variables and multiplying them togethers. See section 11.4.2 of the Stata User's Guide PDF included with your Stata installation and accessible from Stata's Help menu for more on this.

        Comment


        • #5
          Wait a second. I read your post a bit too fast. You are using the treatment as the dependent variable. I assumed you were trying to see if the treatment had some effect on the outcome. However, if this was really your goal, then there's not much more to add. You have shown that few of your measured covariates have any effect on the treatment assignment. In retrospect, you appear to have hand-coded dummy variables for education, income, and age. It is better and more convenient to keep them in one variable, and use the factor variable syntax. If you coded them correctly, including omitting a base category for each of the categorical variables, then I believe the regression results are still correct.

          Everything else I said still applies. There is no correction you can or should apply.

          I maintain that the advice to use a multinomial model for a binary dependent variable is strange, but the model is equivalent to a binary probit model in this case. The output table is just arranged a bit differently.
          Be aware that it can be very hard to answer a question without sample data. You can use the dataex command for this. Type help dataex at the command line.

          When presenting code or results, please use the code delimiters format them. Use the # button on the formatting toolbar, between the " (double quote) and <> buttons.

          Comment


          • #6
            Thank you very much for your reply

            William, My main case is children. gender here refers to child and occupation here refers to father; Could you explain me what you mean by having included employment without the interactions?

            Comment


            • #7
              I'm sorry, your names of your variables were unclear and I thought the "f" before the occupation indicated an interaction with your "female" variable.

              You have not commented that your probit reports only 249 observations, even though you said 2,010 children had diarrhea. That means that 1,761 observations were omitted from your probit because of missing values for your independent variables. You have a huge selection problem there. You need to understand why you have so many missing values.
              Code:
              misstable s614a female two_child many_child ///
              agegrp2 agegrp3 agegrp4 agegrp5 agegrp6 agegrp7 ///
              priedu secedu hgedu fpriedu fsecedu fhgedu ///
              wipoorer wimiddle wiricher wirichest earning v714 ///
              fprofessionls fagr_service fmanual ///
              insured v481a v481b v481c urban ///
              if !missing(no_treatment), showzeros
              should help you find the variables that are most affected by missing values, among the children who have diarrhea (assuming no_treatment is missing for children who did not have diarrhea and thus were not eligible for treatment).
              Last edited by William Lisowski; 11 Apr 2018, 11:44.

              Comment


              • #8
                Hello William,

                Thank you very much for your reply. Sorry for ambiguous writing.

                # of children had diahrrea are 2,010. children did not have diahrrea are 13,456,
                children with 'No treatment' with are1,331 obs. and children had treatment with 679 obs.

                My independent variables are: s614a:- number of diarrhea in last month; female:- Child's sex, # of children:- agegrp2 agegrp3 agegrp4 agegrp5 agegrp6 agegrp7;
                Mother's education:- priedu secedu hgedu; Father's education:- fpriedu fsecedu fhgedu; Household wealth status:- wipoorer wimiddle wiricher wirichest;
                Father's occupation:- fprofessionls fagr_service fmanual; Mother insures:- insured; Insurance Type:- v481a v481b v481c; Area: urban

                I know why I have missing values,
                Your assumption is right, As if the child did not have diarrhea then treatment is missing. Treatment is zero if there is no treatment; Treatment is one if child had treatment
                I do not want to replace the missing values with any other values. I do not want to remove missing values. In my model missing values refers to children did not have diarrhea.

                That's why I am trying to apply Heckprobit and obtain Miles Ratio; hereafter use it as explanatory variable in the second probit model.

                I tried to use the Probit Model instead of Multinomial Model as well as used factor variables; however; there is no big difference appeared in the results and R2 is too small ( 0.1068). Moreover Chi2 is insignificant.

                Much appreciated your comments and Please if you have any further suggestions that would be very welcomed



                Comment


                • #9
                  In post #7 I wrote

                  You have not commented that your probit reports only 249 observations, even though you said 2,010 children had diarrhea.
                  You have not addressed this. In post #8 you again wrote

                  children with 'No treatment' with are1,331 obs. and children had treatment with 679 obs.
                  To which I again must say that you need to address why your probit only shows 249 observations and not 2,010 observations? You have omitted 1,761 observations - 88% of your data - from your probit.

                  To say it another way. Stata omits observations with missing values in the dependent or independent variables. Certainly you want Stata to omit observations with missing values for the dependent variable no_treatment, and that reduces your observations from 15,466 to 2,010. But Stata is also omitting observations with missing values for your independent variables, and that reduces your observations from 2,010 to just 249.

                  I do not want to replace the missing values with any other values. I do not want to remove missing values. In my model missing values refers to children did not have diarrhea.
                  You have missing values in your independent variables and the misstable command I gave will tell you which variables have missing values. Please run it and review the results. Before you worry about selection bias against the children without diarrhea, you must understand why you are ignoring 88% of your children with diarrhea. The missing 88% of your data may explain why your results are so poor.

                  Comment


                  • #10
                    Embarika, once you've read and addressed William's discussion - and you really should! - you may want to review the native Stata command -heckprobit-. Does it do what you need it to? The manual doesn't explicitly discuss the inverse Mills ratio, but I believe it should suffice and save you the trouble of hand-calculating that ratio.

                    As to the issue of multinomial model vs binary model, it could be that the person who advised you thought that people who had no diarrhea were coded as something non-missing. If you want to achieve that goal, you can just replace the missing values as needed, e.g. if s614a > 0. However, you have to address the problem with why Stata is dropping more people than are missing the treatment variable.

                    Clearly, I could have done a bit more to read your second post carefully. However, your first post did not adequately convey what you were trying to do. We cannot help you if you do not help us.
                    Be aware that it can be very hard to answer a question without sample data. You can use the dataex command for this. Type help dataex at the command line.

                    When presenting code or results, please use the code delimiters format them. Use the # button on the formatting toolbar, between the " (double quote) and <> buttons.

                    Comment


                    • #11
                      Hello, Weiwen and William

                      Thank you very much for your help. Really appreciate it.

                      Followed William's advice, I wrote misstable summarize and it found that one of my explanatory variables caused this problem.

                      My goal is to address the determinants of treatment and the choice of treatment providers, However I had children who did not experience diarrhea and therefore, I need to apply Heckman Selection Bias routine to avoid the selection bias issue. I did the routine and it works fine. Here as it appears

                      Code:
                       . heckprobit had_treatment female v218 v133 v715 agegrp2 agegrp3 agegrp4 ///
                      >  agegrp5 agegrp6 agegrp7  ///
                      > wipoorer wimiddle wiricher wirichest v714 fprofessionls fagr_service fmanual ///
                      > i.v025, select(had_diahrrea= improved_sanitation v160 ///
                      > female v218 v133 v715 agegrp2 agegrp3 agegrp4 agegrp5 agegrp6 agegrp7 ///
                      > wipoorer wimiddle wiricher wirichest  v714 ///
                      > fprofessionls fagr_service fmanual i.v025) vce(cluster clusterno)
                      
                      Fitting probit model:
                      
                      Iteration 0:   log pseudolikelihood = -1274.2322 
                      Iteration 1:   log pseudolikelihood = -1260.4879 
                      Iteration 2:   log pseudolikelihood = -1260.4831 
                      Iteration 3:   log pseudolikelihood = -1260.4831 
                      
                      Fitting selection model:
                      
                      Iteration 0:   log pseudolikelihood = -5947.0121 
                      Iteration 1:   log pseudolikelihood = -5593.1371 
                      Iteration 2:   log pseudolikelihood = -5585.8404 
                      Iteration 3:   log pseudolikelihood =   -5585.82 
                      Iteration 4:   log pseudolikelihood =   -5585.82 
                      
                      Fitting starting values:
                      
                      Iteration 0:   log pseudolikelihood = -1385.6012 
                      Iteration 1:   log pseudolikelihood = -1259.0971 
                      Iteration 2:   log pseudolikelihood = -1258.7786 
                      Iteration 3:   log pseudolikelihood = -1258.7786 
                      
                      Fitting full model:
                      
                      Iteration 0:   log pseudolikelihood = -8609.2473 
                      Iteration 1:   log pseudolikelihood = -6907.2012  (not concave)
                      Iteration 2:   log pseudolikelihood = -6852.9199 
                      Iteration 3:   log pseudolikelihood = -6844.6571  (not concave)
                      Iteration 4:   log pseudolikelihood = -6844.6373 
                      Iteration 5:   log pseudolikelihood = -6844.5798 
                      Iteration 6:   log pseudolikelihood = -6844.5056 
                      Iteration 7:   log pseudolikelihood = -6844.4545 
                      Iteration 8:   log pseudolikelihood = -6844.4487 
                      Iteration 9:   log pseudolikelihood = -6844.4487 
                      
                      Probit model with sample selection              Number of obs     =     15,417
                                                                      Censored obs      =     13,418
                                                                      Uncensored obs    =      1,999
                      
                                                                      Wald chi2(19)     =     210.55
                      Log pseudolikelihood = -6844.449                Prob > chi2       =     0.0000
                      
                                                      (Std. Err. adjusted for 11,352 clusters in clusterno)
                      -------------------------------------------------------------------------------------
                                          |               Robust
                                          |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
                      --------------------+----------------------------------------------------------------
                      had_treatment       |
                                   female |  -.0984126   .0434199    -2.27   0.023     -.183514   -.0133113
                                     v218 |  -.0449089   .0167034    -2.69   0.007     -.077647   -.0121708
                                     v133 |  -.0107149   .0061958    -1.73   0.084    -.0228585    .0014286
                                     v715 |  -.0003573   .0050923    -0.07   0.944     -.010338    .0096234
                                  agegrp2 |   .3049425   .1008603     3.02   0.002       .10726    .5026251
                                  agegrp3 |   .4165191   .0893854     4.66   0.000     .2413269    .5917114
                                  agegrp4 |   .2389965   .0841835     2.84   0.005     .0739998    .4039932
                                  agegrp5 |  -.0446871   .0882475    -0.51   0.613     -.217649    .1282748
                                  agegrp6 |  -.2219974   .1166897    -1.90   0.057     -.450705    .0067103
                                  agegrp7 |   -.556972   .1080802    -5.15   0.000    -.7688053   -.3451387
                                 wipoorer |  -.0370999    .068264    -0.54   0.587    -.1708949    .0966951
                                 wimiddle |   -.104658     .06831    -1.53   0.125    -.2385431    .0292271
                                 wiricher |  -.0121352   .0893742    -0.14   0.892    -.1873054    .1630349
                                wirichest |  -.2020356   .0966176    -2.09   0.037    -.3914026   -.0126686
                                     v714 |   .1375346    .066711     2.06   0.039     .0067835    .2682857
                            fprofessionls |  -.1712724   .1596783    -1.07   0.283    -.4842362    .1416914
                             fagr_service |  -.1173784   .1576626    -0.74   0.457    -.4263914    .1916346
                                  fmanual |   -.064183   .1573216    -0.41   0.683    -.3725276    .2441617
                                          |
                                     v025 |
                                   rural  |   .0460665   .0657959     0.70   0.484    -.0828911    .1750242
                                    _cons |   -.846423   .3350136    -2.53   0.012    -1.503038   -.1898085
                      --------------------+----------------------------------------------------------------
                      had_diahrrea        |
                      improved_sanitation |   .0451389   .0322723     1.40   0.162    -.0181135    .1083914
                                     v160 |  -.0375564   .0123564    -3.04   0.002    -.0617746   -.0133382
                                   female |  -.0441726   .0266826    -1.66   0.098    -.0964694    .0081243
                                     v218 |  -.0324522   .0111602    -2.91   0.004    -.0543257   -.0105787
                                     v133 |  -.0155404   .0036889    -4.21   0.000    -.0227704   -.0083103
                                     v715 |  -.0001738   .0031008    -0.06   0.955    -.0062513    .0059038
                                  agegrp2 |   .2735892   .0717935     3.81   0.000     .1328765    .4143019
                                  agegrp3 |   .4194442   .0616261     6.81   0.000     .2986593     .540229
                                  agegrp4 |   .2499122   .0580507     4.31   0.000     .1361349    .3636895
                                  agegrp5 |  -.0791277   .0596034    -1.33   0.184    -.1959482    .0376928
                                  agegrp6 |  -.3442093   .0613154    -5.61   0.000    -.4643853   -.2240334
                                  agegrp7 |  -.5515667   .0685205    -8.05   0.000    -.6858645   -.4172689
                                 wipoorer |  -.1055529   .0446514    -2.36   0.018     -.193068   -.0180378
                                 wimiddle |  -.1691474   .0468217    -3.61   0.000    -.2609163   -.0773785
                                 wiricher |  -.1489741   .0556476    -2.68   0.007    -.2580413   -.0399068
                                wirichest |  -.2468304    .070063    -3.52   0.000    -.3841514   -.1095094
                                     v714 |   .0792096   .0441871     1.79   0.073    -.0073956    .1658147
                            fprofessionls |  -.1038408   .1096902    -0.95   0.344    -.3188296    .1111481
                             fagr_service |  -.1206469   .1102402    -1.09   0.274    -.3367138      .09542
                                  fmanual |   .0061716   .1077484     0.06   0.954    -.2050115    .2173546
                                          |
                                     v025 |
                                   rural  |   .0584437   .0454868     1.28   0.199    -.0307088    .1475963
                                    _cons |   -.756077   .1339762    -5.64   0.000    -1.018666   -.4934884
                      --------------------+----------------------------------------------------------------
                                  /athrho |   1.464254   .7687172     1.90   0.057    -.0424043    2.970912
                      --------------------+----------------------------------------------------------------
                                      rho |   .8984756   .1481638                     -.0423789    .9947593
                      -------------------------------------------------------------------------------------
                      Wald test of indep. eqns. (rho = 0): chi2(1) =     3.63   Prob > chi2 = 0.0568
                      
                      .
                      end of do-file
                      
                      .
                      Because my dependent variables in both the selection equation and main equation are binary; I need to use Heckprobit command instead of Heckman command. Through Heckprobit command does not give you directly lambdas (the inverse Mills ratio), you need to calculate it!

                      In my dataset, if the child did not have diahrrea, correspondingly it reflects a missing value in the treatment variable. Treatment variable coded as is zero if there is no treatment; is one if child had treatment. As well as it is ' . ' if the children did not have diarrhea.

                      I tried to use factor variable and use age as continuous variable instead of many binary variables, still the results are not improved. Please see below:

                      Code:
                       . probit no_treatment s614a female v218 b8 ///
                      > i.v106 i.v701 ///
                      > i.v190  v714 ///
                      > fprofessionls fagr_service fmanual ///
                      > insured v481a v481b v481c urban
                      
                      Iteration 0:   log likelihood = -1278.7266 
                      Iteration 1:   log likelihood = -1255.7238 
                      Iteration 2:   log likelihood = -1255.6844 
                      Iteration 3:   log likelihood = -1255.6844 
                      
                      Probit regression                               Number of obs     =      2,000
                                                                      LR chi2(23)       =      46.08
                                                                      Prob > chi2       =     0.0029
                      Log likelihood = -1255.6844                     Pseudo R2         =     0.0180
                      
                      -------------------------------------------------------------------------------
                       no_treatment |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
                      --------------+----------------------------------------------------------------
                              s614a |   .0913016   .0250447     3.65   0.000     .0422149    .1403884
                             female |   -.102957   .0587461    -1.75   0.080    -.2180972    .0121832
                               v218 |  -.0351586   .0230811    -1.52   0.128    -.0803966    .0100794
                                 b8 |  -.0536658   .0245884    -2.18   0.029    -.1018582   -.0054734
                                    |
                               v106 |
                           primary  |  -.0240806    .118754    -0.20   0.839    -.2568342    .2086731
                         secondary  |   .0986642   .0891344     1.11   0.268    -.0760361    .2733645
                            higher  |  -.0496741   .1398262    -0.36   0.722    -.3237284    .2243802
                                    |
                               v701 |
                           primary  |    .021242    .115198     0.18   0.854     -.204542     .247026
                         secondary  |   .0192647   .0976591     0.20   0.844    -.1721436     .210673
                            higher  |  -.1194293   .1398771    -0.85   0.393    -.3935834    .1547248
                                    |
                               v190 |
                            poorer  |   .0644305   .0897046     0.72   0.473    -.1113872    .2402483
                            middle  |   .0071112   .0925451     0.08   0.939    -.1742739    .1884962
                            richer  |   .1087628   .1084571     1.00   0.316    -.1038092    .3213348
                           richest  |   -.103245   .1389801    -0.74   0.458     -.375641    .1691509
                                    |
                               v714 |   .1100308    .102338     1.08   0.282     -.090548    .3106096
                      fprofessionls |  -.1264724    .220164    -0.57   0.566     -.557986    .3050412
                       fagr_service |  -.0672151   .2197128    -0.31   0.760    -.4978443    .3634141
                            fmanual |  -.1518738   .2145235    -0.71   0.479    -.5723322    .2685846
                            insured |  -.0954954   .6103683    -0.16   0.876    -1.291795    1.100805
                              v481a |  -.0420414   .6281507    -0.07   0.947    -1.273194    1.189111
                              v481b |   .3216119   .6158674     0.52   0.602     -.885466     1.52869
                              v481c |     .57041   .7565891     0.75   0.451    -.9124775    2.053297
                              urban |    .025002    .094685     0.26   0.792    -.1605772    .2105813
                              _cons |   .4721861   .2547669     1.85   0.064    -.0271479    .9715201
                      -------------------------------------------------------------------------------
                      
                      .
                      end of do-file
                      
                      .

                      Comment

                      Working...
                      X