Count model with data panel ( poisson)

[Kevin Hoyos]

Hi, I'm new here (and new to stata/statistical analysis) i need help with a count model with data panel.

i have some panel from a count of a financial transactions per transactional channel of 500 customers of a financial institution , i try to evaluate the impact of diferent socioeconomics variebles in the demand of each channel.
As my dataset contains many many 0 (there are no transactions in a channel i for a client j), I understood that maybe a poisson panel model were appropriate in my case, to account for 0 and overdispersion
So I tried to estimate
xtpoisson txn_canal multicanal genero edad activos pasivos tarjetas
i try to estimate marginals effects but i can't use " margins"
Why my independent variables is "not found in the list of covariates"
what can i do?
Thanks you for your help!!

[Joro Kolev]

At the very minimum post the exact commands you executed, and the exact error messages that Stata returned.

You might also want to check -dataex- and provide a minimal data example that generates the problem.

[Joro Kolev]

And when you are posting your commands and Stata output, use appropriate formatting, see the sharp sign (#) at the formatting line of your post , it will give you properly formatted output and commands.

[Kevin Hoyos]

At the very minimum post the exact commands you executed, and the exact error messages that Stata returned.

You might also want to check -dataex- and provide a minimal data example that generates the problem.

sorry i`m new in this , the exact comand is:



xtpoisson txn_canal multicanal genero edad activos pasivos tarjetas

Fitting Poisson model:

Iteration 0: log likelihood = -475048.27
Iteration 1: log likelihood = -470630.91
Iteration 2: log likelihood = -470602.32
Iteration 3: log likelihood = -470602.31

Fitting full model:

Iteration 0: log likelihood = -320820.94
Iteration 1: log likelihood = -320805.85
Iteration 2: log likelihood = -320805.56
Iteration 3: log likelihood = -320805.56

Random-effects Poisson regression Number of obs = 2,821
Group variable: num_doc_org Number of groups = 403

Random effects u_i ~ Gamma Obs per group:
min = 7
avg = 7.0
max = 7

Wald chi2(6) = 132.32
Log likelihood = -320805.56 Prob > chi2 = 0.0000

------------------------------------------------------------------------------
txn_canal | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
multicanal | 1.759168 .4919221 3.58 0.000 .7950184 2.723318
genero | .3843714 .1109466 3.46 0.001 .16692 .6018228
edad | -.0100879 .0039324 -2.57 0.010 -.0177953 -.0023805
activos | .4855768 .114347 4.25 0.000 .2614607 .7096929
pasivos | .4166244 .1179023 3.53 0.000 .1855401 .6477086
tarjetas | .4183876 .0625 6.69 0.000 .2958898 .5408853
_cons | 2.616826 .2714788 9.64 0.000 2.084737 3.148914
-------------+----------------------------------------------------------------
/lnalpha | .1811987 .0612294 .0611913 .3012061
-------------+----------------------------------------------------------------
alpha | 1.198653 .0733928 1.063102 1.351488
------------------------------------------------------------------------------
LR test of alpha=0: chibar2(01) = 3.0e+05 Prob >= chibar2 = 0.000

margins genero edad activos pasivos tarjetas
factor 'genero' not found in list of covariates
r(322);

[Joro Kolev]

These variables sound to me like continuous variables.

Is not what you need:

margins, dydx(*)

Like in

. webuse ships, clear

. xtset ship
       panel variable:  ship (balanced)

. xtpoisson accident service, nolog

Random-effects Poisson regression               Number of obs     =         34
Group variable: ship                            Number of groups  =          5

Random effects u_i ~ Gamma                      Obs per group:
                                                              min =          6
                                                              avg =        6.8
                                                              max =          7

                                                Wald chi2(1)      =      20.18
Log likelihood  = -137.36653                    Prob > chi2       =     0.0000

------------------------------------------------------------------------------
    accident |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
     service |   .0000208   4.63e-06     4.49   0.000     .0000117    .0000299
       _cons |   2.038602    .404576     5.04   0.000     1.245647    2.831556
-------------+----------------------------------------------------------------
    /lnalpha |  -.2464067    .589829                      -1.40245    .9096369
-------------+----------------------------------------------------------------
       alpha |   .7816043   .4610129                      .2459935    2.483421
------------------------------------------------------------------------------
LR test of alpha=0: chibar2(01) = 132.29               Prob >= chibar2 = 0.000

. margins, dydx(*)

Average marginal effects                        Number of obs     =         34
Model VCE    : OIM

Expression   : Linear prediction, predict()
dy/dx w.r.t. : service

------------------------------------------------------------------------------
             |            Delta-method
             |      dy/dx   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
     service |   .0000208   4.63e-06     4.49   0.000     .0000117    .0000299
------------------------------------------------------------------------------

.

[Kevin Hoyos]

These variables sound to me like continuous variables.

Is not what you need:

margins, dydx(*)

Like in

. webuse ships, clear

. xtset ship
panel variable: ship (balanced)

. xtpoisson accident service, nolog

Random-effects Poisson regression Number of obs = 34
Group variable: ship Number of groups = 5

Random effects u_i ~ Gamma Obs per group:
min = 6
avg = 6.8
max = 7

Wald chi2(1) = 20.18
Log likelihood = -137.36653 Prob > chi2 = 0.0000

------------------------------------------------------------------------------
accident | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
service | .0000208 4.63e-06 4.49 0.000 .0000117 .0000299
_cons | 2.038602 .404576 5.04 0.000 1.245647 2.831556
-------------+----------------------------------------------------------------
/lnalpha | -.2464067 .589829 -1.40245 .9096369
-------------+----------------------------------------------------------------
alpha | .7816043 .4610129 .2459935 2.483421
------------------------------------------------------------------------------
LR test of alpha=0: chibar2(01) = 132.29 Prob >= chibar2 = 0.000

. margins, dydx(*)

Average marginal effects Number of obs = 34
Model VCE : OIM

Expression : Linear prediction, predict()
dy/dx w.r.t. : service

------------------------------------------------------------------------------
| Delta-method
| dy/dx Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
service | .0000208 4.63e-06 4.49 0.000 .0000117 .0000299
------------------------------------------------------------------------------

.

if they are continuous variables .
Thanks very much you for your help, was very helpful!