[Denis Beuca]

So for example : the Random model predicts that on average females have a higher wellbeing of 96.2% compared to men holding all other variables constant ?

[Joro Kolev]

So for example : the Random model predicts that on average females have a higher wellbeing of 96.2% compared to men holding all other variables constant ?

No, the average man has 0.962 higher wellbeing on whatever scale you are measuring wellbeing, compared to the average woman.

[Denis Beuca]

Thanks for the clarification just to further confirm in the case for my Children1 variable 0 = no kids and 1= one or more kids, the coefficient in the random model is 0.0148 so in this instance the average person who has no kids has a 0.0148 higher wellbeing than someone who has kids ?

[Clyde Schechter]

No, it's the other way around.

[Joro Kolev]

And if you are interested in estimating the effect of such things as gender, you cannot use the person specific fixed effects model, because the majority of people do not change their gender, and therefore gender will be collinear with the person fixed effects.

[Carlo Lazzaro]

xtreg wellbeing lnincome gender children1 i.economicactivity i.education t if balanced==1, vce(cluster pidp)

I'm not sure how to show what Stata gave me back

Denis:
just put everything in between CODE delimiters.

[Denis Beuca]

xtreg wellbeing lnincome gender children1 i.economicactivity i.education t if balanced==1, vce(cluster pidp )

[Denis Beuca]

 xtreg wellbeing lnincome gender children1 i.economicactivity i.education t if balanced==1, vce(cluster pidp )

Random-effects GLS regression                   Number of obs     =     80,616
Group variable: pidp                            Number of groups  =     10,077

R-squared:                                      Obs per group:
     Within  = 0.0037                                         min =          8
     Between = 0.0798                                         avg =        8.0
     Overall = 0.0433                                         max =          8

                                                Wald chi2(10)     =     639.88
corr(u_i, X) = 0 (assumed)                      Prob > chi2       =     0.0000

                                               (Std. err. adjusted for 10,077 clusters in pidp)
-----------------------------------------------------------------------------------------------
                              |               Robust
                    wellbeing | Coefficient  std. err.      z    P>|z|     [95% conf. interval]
------------------------------+----------------------------------------------------------------
                     lnincome |   .0354959   .0287192     1.24   0.216    -.0207927    .0917845
                       gender |   .9620263    .074427    12.93   0.000      .816152    1.107901
                    children1 |   .0148389    .069374     0.21   0.831    -.1211316    .1508095
                              |
             economicactivity |
                  unemployed  |  -1.777649   .1698494   -10.47   0.000    -2.110548   -1.444751
                     retired  |   .6889726   .0640266    10.76   0.000     .5634828    .8144625
                    students  |  -.2966172   .2671824    -1.11   0.267    -.8202851    .2270507
                       other  |   -1.43992   .1194814   -12.05   0.000    -1.674099    -1.20574
                              |
                    education |
Alevel or GCSE or Other qual  |  -.1517329   .0742707    -2.04   0.041    -.2973008   -.0061651
                     No qual  |  -.3851673   .1486346    -2.59   0.010    -.6764858   -.0938488
                              |
                            t |  -.0453144   .0071014    -6.38   0.000    -.0592329   -.0313958
                        _cons |    24.7853   .2222687   111.51   0.000     24.34966    25.22094
------------------------------+----------------------------------------------------------------
                      sigma_u |  3.3699175
                      sigma_e |  3.6862565
                          rho |  .45525841   (fraction of variance due to u_i)
-----------------------------------------------------------------------------------------------

.

[Carlo Lazzaro]

Denis:
1) what does the -xttest0- outcome tell you?
2) the between Rsq seems a tad low. I'd double-check if the functional form of the regressand is correctly specified;
3) focusing on the balanced subsample only, may give you results that are unreliable when the data generating process is taken into account.

[Denis Beuca]

Breusch and Pagan Lagrangian multiplier test for random effects

        wellbeing[pidp,t] = Xb + u[pidp] + e[pidp,t]

        Estimated results:
                         |       Var     SD = sqrt(Var)
                ---------+-----------------------------
               wellbeing |   26.45754        5.14369
                       e |   13.58849       3.686256
                       u |   11.35634       3.369918

        Test: Var(u) = 0
                             chibar2(01) = 57994.23
                          Prob > chibar2 =   0.0000

[Denis Beuca]

I have a further question when i try to restrict the sample to males only i get an error saying invalid syntax
xtreg wellbeing lnincome gender children1 i.economicactivity i.education t if balanced==1 if gender==1, fe vce(cluster pidp ) I would really appreciate any help in regards to this

[Joro Kolev]

I have a further question when i try to restrict the sample to males only i get an error saying invalid syntax
xtreg wellbeing lnincome gender children1 i.economicactivity i.education t if balanced==1 if gender==1, fe vce(cluster pidp ) I would really appreciate any help in regards to this

Try instead

xtreg wellbeing lnincome gender children1 i.economicactivity i.education t if balanced==1 & gender==1, fe vce(cluster pidp )

as I told you before, everything that does not change across time will be dropped.

[Carlo Lazzaro]

Breusch and Pagan Lagrangian multiplier test for random effects

wellbeing[pidp,t] = Xb + u[pidp] + e[pidp,t]

Estimated results:
| Var SD = sqrt(Var)
---------+-----------------------------
wellbeing | 26.45754 5.14369
e | 13.58849 3.686256
u | 11.35634 3.369918

Test: Var(u) = 0
chibar2(01) = 57994.23
Prob > chibar2 = 0.0000

Denis:
the -xttest0- outcome supports the evidence of a panel-wise effect.