DIfferent individuals in one merged data set

[Kasia Kiki]

Hi all,
I have a problem, because I have already merged data with ID numbers from different waves of survey into one merged data-set. I asked the statistical center and they replied me, that individuals were not the same across years but nevertheless they got the same number in a merged data set.. Therefore which panel data regression I can use? I excluded fixed effects, because individuals are different over time. Can i Use random effects and which explanation I can write that i used it?
All the best,

[George Ford]

do you have an indicator from which data set the observations came from? if so, you can create an id based on that.

egen id = group(obsid, dataset)

[George Ford]

oops. no comma in group

[Kasia Kiki]

No, just all IDs are merged into one and is it a huge mistake if I treat them for the purpose of model as same individuals? ( they were choosen randomly in one, same area but just different individuals)

[Andrew Musau]

My reading of #1 is that you are asking about modeling options and not how to create identifiers. You do not have panel data but pooled cross-sections, so rule out panel data models (FE, RE). Thus, you are confined to linear regression assuming a continuous outcome, but you may include time effects. If you want the coefficient on a variable to vary across time, interact the variable with the time dummies. You will still have the problem of unobserved heterogeneity that is endemic in cross-sectional analyses.

[Kasia Kiki]

yes exactly, Thank you very much, therefore instead of (xtreg) should I just use (reg) command? should I also set data as panel data (xtset) or how would teh commands look like if i perform pooled OLS?

[Andrew Musau]

You can use either regress or xtreg, fe in case you include time effects. With regress, you include time dummies and there is no need for xtset whereas with xtreg, fe, you xtset using the time variable. If we suppose that the Grunfeld dataset consists of pooled cross-sections, the following are equivalent:

webuse grunfeld, clear
gen id=_n
xtset year
xtreg invest mvalue kstock, fe cluster(id) nonest dfadj
regress invest mvalue kstock i.year, robust

Clustering at the observation level yields White standard errors in xtreg.

Res.:

. xtreg invest mvalue kstock, fe cluster(id) nonest dfadj

Fixed-effects (within) regression               Number of obs      =       200
Group variable: year                            Number of groups   =        20

R-sq:  within  = 0.8038                         Obs per group: min =        10
       between = 0.9325                                        avg =      10.0
       overall = 0.8122                                        max =        10

                                                F(2,178)           =    192.01
corr(u_i, Xb)  = 0.0565                         Prob > F           =    0.0000

                                   (Std. Err. adjusted for 200 clusters in id)
------------------------------------------------------------------------------
             |               Robust
      invest |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      mvalue |   .1167978   .0066825    17.48   0.000     .1036107    .1299849
      kstock |   .2197066   .0575203     3.82   0.000     .1061972     .333216
       _cons |   -41.0225   13.59568    -3.02   0.003    -67.85195   -14.19304
-------------+----------------------------------------------------------------
     sigma_u |  15.309436
     sigma_e |  98.099115
         rho |  .02377594   (fraction of variance due to u_i)
------------------------------------------------------------------------------

. 
. regress invest mvalue kstock i.year, robust

Linear regression                               Number of obs     =        200
                                                F(21, 178)        =      23.74
                                                Prob > F          =     0.0000
                                                R-squared         =     0.8170
                                                Root MSE          =     98.099

------------------------------------------------------------------------------
             |               Robust
      invest |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      mvalue |   .1167978   .0066825    17.48   0.000     .1036107    .1299849
      kstock |   .2197066   .0575203     3.82   0.000     .1061972     .333216
             |
        year |
       1936  |  -17.21234   32.30392    -0.53   0.595    -80.96027    46.53559
       1937  |  -34.49127   39.41671    -0.88   0.383    -112.2755    43.29291
       1938  |  -28.44276   26.44632    -1.08   0.284    -80.63142    23.74589
       1939  |  -56.24304   30.29841    -1.86   0.065    -116.0334     3.54727
       1940  |  -30.50473   26.88752    -1.13   0.258    -83.56404    22.55459
       1941  |  -2.627113   27.07946    -0.10   0.923    -56.06521    50.81098
       1942  |  -1.422156   27.25112    -0.05   0.958      -55.199    52.35469
       1943  |  -21.80127   26.90696    -0.81   0.419    -74.89895    31.29641
       1944  |  -22.11735   25.50119    -0.87   0.387    -72.44091    28.20622
       1945  |  -33.59647   24.79854    -1.35   0.177    -82.53343     15.3405
       1946  |  -7.028064   31.03571    -0.23   0.821    -68.27333     54.2172
       1947  |  -5.246123   31.18714    -0.17   0.867    -66.79023    56.29799
       1948  |  -3.919472   39.34635    -0.10   0.921    -81.56481    73.72587
       1949  |  -28.79332   35.41855    -0.81   0.417    -98.68761    41.10098
       1950  |  -28.35409   38.03333    -0.75   0.457    -103.4083    46.70015
       1951  |  -11.67194   45.29242    -0.26   0.797    -101.0511    77.70727
       1952  |  -5.613218   50.98644    -0.11   0.912    -106.2289    95.00245
       1953  |   2.448996   52.49263     0.05   0.963     -101.139     106.037
       1954  |  -12.31488   41.68043    -0.30   0.768    -94.56625     69.9365
             |
       _cons |  -23.57497   14.25381    -1.65   0.100    -51.70316    4.553223
------------------------------------------------------------------------------

.