Hi all! I'm new to the forum and I'm coming here seeking some advice. I'm currently trying to estimate the effect of remote work on worklife balance and well being of mothers with fixed effects and I'm having a hard time deciding what to control to when it comes to occupation/industry.
The issue is the following: the occupation variable as is, contains 9330 categories with a total of 26178 valid observations in my dataset. (Link to the variable details: https://paneldata.org/soep-core/data...equiv/e1110598 ).
I also have the choice to use the 1- digit industry code (10 categories, 23971 observations) ( https://paneldata.org/soep-core/data...quiv/e1110697 ) and 2-digit industry code ( 33 categories, 23971 observations) (https://paneldata.org/soep-core/data...equiv/e1110797 )
I ended up picking the occupation variable, as it gave me the highest within R-squared (0.0903 within R-squared with occupation variable versus 0.0434 within R-squared with the 2-digit industry code) ; however only now I have realized that it the occupation variable basically serves as 9330 dummy variables, and because of that the sheer number alone could've accounted for the doubling of the R-squared (This is my understanding, please correct me if I'm wrong). Should I continue to only use occupation variable, or switch and just settle with only having an industry 1 or 2 digit code?
Thank you and kind regards,
Matej
Code with occupational variable (I've cut most of the +9000 occupational dummies due to length)
Code with 2-digit industry code:
Code with 1-digi industry code
The issue is the following: the occupation variable as is, contains 9330 categories with a total of 26178 valid observations in my dataset. (Link to the variable details: https://paneldata.org/soep-core/data...equiv/e1110598 ).
I also have the choice to use the 1- digit industry code (10 categories, 23971 observations) ( https://paneldata.org/soep-core/data...quiv/e1110697 ) and 2-digit industry code ( 33 categories, 23971 observations) (https://paneldata.org/soep-core/data...equiv/e1110797 )
I ended up picking the occupation variable, as it gave me the highest within R-squared (0.0903 within R-squared with occupation variable versus 0.0434 within R-squared with the 2-digit industry code) ; however only now I have realized that it the occupation variable basically serves as 9330 dummy variables, and because of that the sheer number alone could've accounted for the doubling of the R-squared (This is my understanding, please correct me if I'm wrong). Should I continue to only use occupation variable, or switch and just settle with only having an industry 1 or 2 digit code?
Thank you and kind regards,
Matej
Code with occupational variable (I've cut most of the +9000 occupational dummies due to length)
Code:
. xtreg overtimehours ib5.freqWFH4##i.children_in_hh_dummy07 i.isced_edu i.syear i.maritalstatus age agesq i.emp
> status i.disability_status logindincome workingexperience i.jobchange2 i.regtyp i.sizecompany firmtime firmtim
> esq i.partnerWFH i.occupation if sex==2 & children_in_hh_dummy816==0 , fe vce(cluster pid)
note: age omitted because of collinearity.
note: 1312.occupation omitted because of collinearity.
note: 1319.occupation omitted because of collinearity.
note: 2121.occupation omitted because of collinearity.
note: 2213.occupation omitted because of collinearity.
note: 2431.occupation omitted because of collinearity.
note: 3232.occupation omitted because of collinearity.
note: 3414.occupation omitted because of collinearity.
note: 7215.occupation omitted because of collinearity.
note: 7324.occupation omitted because of collinearity.
note: 7332.occupation omitted because of collinearity.
note: 7423.occupation omitted because of collinearity.
note: 7442.occupation omitted because of collinearity.
note: 8122.occupation omitted because of collinearity.
note: 8240.occupation omitted because of collinearity.
note: 8266.occupation omitted because of collinearity.
note: 8312.occupation omitted because of collinearity.
note: 9330.occupation omitted because of collinearity.
Fixed-effects (within) regression Number of obs = 8,339
Group variable: pid Number of groups = 3,884
R-squared: Obs per group:
Within = 0.0903 min = 1
Between = 0.0286 avg = 2.1
Overall = 0.0363 max = 5
F(188, 3883) = .
corr(u_i, Xb) = -0.4376 Prob > F = .
(Std. err. adjusted for 3,884 clusters in pid)
---------------------------------------------------------------------------------------------------------------
| Robust
overtimehours | Coefficient std. err. t P>|t| [95% conf. interval]
----------------------------------------------+----------------------------------------------------------------
freqWFH4 |
Daily | 1.580801 .7068852 2.24 0.025 .1948997 2.966703
Semi-frequent, at least monthly | .99053 .5133477 1.93 0.054 -.0159267 1.996987
|
1.children_in_hh_dummy07 | -.9140177 .2526063 -3.62 0.000 -1.409271 -.4187641
|
freqWFH4#children_in_hh_dummy07 |
Daily#1 | -1.47619 1.292971 -1.14 0.254 -4.011156 1.058776
Semi-frequent, at least monthly#1 | -2.155114 .9060992 -2.38 0.017 -3.931589 -.3786379
|
isced_edu |
intermediate edu | .8347827 .3438065 2.43 0.015 .1607242 1.508841
higher edu | 1.295067 .4511599 2.87 0.004 .4105343 2.1796
|
syear |
1999 | .3112715 .3356538 0.93 0.354 -.3468029 .9693459
2002 | .7115967 .3298794 2.16 0.031 .0648434 1.35835
2009 | 2.049542 .7357149 2.79 0.005 .6071182 3.491967
2014 | 2.34986 1.048949 2.24 0.025 .2933169 4.406404
|
maritalstatus |
Married, But Separated | .2121298 .3736198 0.57 0.570 -.5203799 .9446395
Single | .1504195 .2267014 0.66 0.507 -.2940455 .5948846
Divorced | .1304869 .3616733 0.36 0.718 -.5786008 .8395746
Widowed | -.422527 .5129552 -0.82 0.410 -1.428214 .5831603
Registered same sex partnership | 1.881282 .2077762 9.05 0.000 1.473921 2.288643
Registered same sex partnership, but separ.. | -1.513691 .7524476 -2.01 0.044 -2.988921 -.0384607
|
age | 0 (omitted)
agesq | -.0014598 .0006405 -2.28 0.023 -.0027156 -.0002041
|
empstatus |
Regular Part-Time Employment | -.4651567 .2181483 -2.13 0.033 -.8928529 -.0374606
1.disability_status | .1299819 .3198697 0.41 0.685 -.4971467 .7571106
logindincome | .0958683 .0873129 1.10 0.272 -.0753151 .2670518
workingexperience | -.0628805 .0241019 -2.61 0.009 -.1101341 -.0156269
|
jobchange2 |
Yes, changed job in last year | -.2996533 .201546 -1.49 0.137 -.6947993 .0954928
|
regtyp |
[2] Rural regions | .2161984 .4080876 0.53 0.596 -.583888 1.016285
|
sizecompany |
2 | .3151676 .2366586 1.33 0.183 -.1488194 .7791546
3 | .2373861 .2621315 0.91 0.365 -.2765425 .7513146
4 | .2590375 .2525779 1.03 0.305 -.2361605 .7542355
5 | .2116392 .2584372 0.82 0.413 -.2950464 .7183248
6 | .4608767 .2867828 1.61 0.108 -.1013825 1.023136
7 | -.0050526 .3459975 -0.01 0.988 -.6834067 .6733015
Unknown | .0747334 .3874861 0.19 0.847 -.6849622 .8344291
|
firmtime | -.0069339 .0253417 -0.27 0.784 -.0566182 .0427505
firmtimesq | .0004206 .0007693 0.55 0.585 -.0010877 .001929
|
partnerWFH |
Partner Working Remotely | .1729584 .2048384 0.84 0.399 -.2286428 .5745595
|
occupation |
1140 | 2.003218 1.163329 1.72 0.085 -.2775754 4.284012
1142 | 1.869942 1.166629 1.60 0.109 -.4173218 4.157205
1200 | .8113705 2.558319 0.32 0.751 -4.204406 5.827147
1210 | .
...................
.........................................................
9161 | .8655941 1.293727 0.67 0.503 -1.670854 3.402043
9211 | -.8618563 1.494222 -0.58 0.564 -3.79139 2.067678
9320 | .2067987 .9640401 0.21 0.830 -1.683274 2.096872
9330 | 0 (omitted)
|
_cons | 2.287602 1.834321 1.25 0.212 -1.308722 5.883926
----------------------------------------------+----------------------------------------------------------------
sigma_u | 3.7081101
sigma_e | 2.9130295
rho | .61837519 (fraction of variance due to u_i)
---------------------------------------------------------------------------------------------------------------
Code with 2-digit industry code:
Code:
. xtreg overtimehours ib5.freqWFH4##i.children_in_hh_dummy07 i.isced_edu i.syear i.maritalstatus age agesq i.emp
> status i.disability_status logindincome workingexperience i.jobchange2 i.regtyp i.sizecompany firmtime firmtim
> esq i.partnerWFH i.industrycode2 if sex==2 & children_in_hh_dummy816==0 , fe vce(cluster pid)
note: age omitted because of collinearity.
Fixed-effects (within) regression Number of obs = 7,765
Group variable: pid Number of groups = 3,737
R-squared: Obs per group:
Within = 0.0434 min = 1
Between = 0.0246 avg = 2.1
Overall = 0.0254 max = 5
F(59, 3736) = .
corr(u_i, Xb) = -0.3463 Prob > F = .
(Std. err. adjusted for 3,737 clusters in pid)
---------------------------------------------------------------------------------------------------------------
| Robust
overtimehours | Coefficient std. err. t P>|t| [95% conf. interval]
----------------------------------------------+----------------------------------------------------------------
freqWFH4 |
Daily | 2.52021 .7570638 3.33 0.001 1.035912 4.004509
Semi-frequent, at least monthly | 1.332821 .5584959 2.39 0.017 .2378345 2.427808
|
1.children_in_hh_dummy07 | -.8797856 .2582233 -3.41 0.001 -1.386058 -.3735133
|
freqWFH4#children_in_hh_dummy07 |
Daily#1 | -1.724419 1.332334 -1.29 0.196 -4.336593 .887754
Semi-frequent, at least monthly#1 | -2.349489 .9344255 -2.51 0.012 -4.181523 -.5174555
|
isced_edu |
intermediate edu | .7932312 .3587109 2.21 0.027 .0899429 1.49652
higher edu | 1.20516 .4579732 2.63 0.009 .3072578 2.103061
|
syear |
1999 | .0557046 .3670551 0.15 0.879 -.6639433 .7753525
2002 | .8134055 .3518604 2.31 0.021 .1235483 1.503263
2009 | 1.979086 .7789014 2.54 0.011 .4519729 3.5062
2014 | 2.242524 1.11455 2.01 0.044 .0573377 4.42771
|
maritalstatus |
Married, But Separated | .2274815 .4186404 0.54 0.587 -.5933045 1.048268
Single | .0327204 .2330377 0.14 0.888 -.4241731 .4896139
Divorced | .1382372 .3800094 0.36 0.716 -.6068089 .8832832
Widowed | -.4293263 .5299967 -0.81 0.418 -1.468438 .6097848
Registered same sex partnership | 1.949983 .214771 9.08 0.000 1.528903 2.371063
Registered same sex partnership, but separ.. | -1.593916 .5594292 -2.85 0.004 -2.690732 -.4970996
|
age | 0 (omitted)
agesq | -.0012855 .000684 -1.88 0.060 -.0026265 .0000555
|
empstatus |
Regular Part-Time Employment | -.49855 .2211557 -2.25 0.024 -.9321478 -.0649523
1.disability_status | -.0180545 .3011389 -0.06 0.952 -.6084672 .5723582
logindincome | .2084391 .0906523 2.30 0.022 .0307063 .3861719
workingexperience | -.0520858 .0253048 -2.06 0.040 -.1016983 -.0024733
|
jobchange2 |
Yes, changed job in last year | -.1938162 .2089521 -0.93 0.354 -.6034876 .2158552
|
regtyp |
[2] Rural regions | .1276912 .422667 0.30 0.763 -.7009894 .9563717
|
sizecompany |
2 | .2236825 .2463555 0.91 0.364 -.2593218 .7066869
3 | .2056656 .2693058 0.76 0.445 -.3223352 .7336664
4 | .1634258 .2591328 0.63 0.528 -.3446297 .6714814
5 | .2753488 .2639699 1.04 0.297 -.2421904 .792888
6 | .5658791 .2876042 1.97 0.049 .0020025 1.129756
7 | .0813208 .3483649 0.23 0.815 -.6016831 .7643246
Unknown | .2781849 .413024 0.67 0.501 -.5315897 1.087959
|
firmtime | -.0277454 .0261347 -1.06 0.288 -.078985 .0234943
firmtimesq | .000589 .0007862 0.75 0.454 -.0009523 .0021304
|
partnerWFH |
Partner Working Remotely | .2272388 .2121365 1.07 0.284 -.1886759 .6431536
|
industrycode2 |
3 | -.4429973 1.534138 -0.29 0.773 -3.450827 2.564833
4 | .8943711 1.833623 0.49 0.626 -2.700629 4.489372
5 | -.5896647 .9113456 -0.65 0.518 -2.376448 1.197119
6 | -.7362645 1.275964 -0.58 0.564 -3.237919 1.76539
7 | -.3418449 .9374674 -0.36 0.715 -2.179843 1.496153
8 | -.1244317 .91229 -0.14 0.892 -1.913067 1.664203
9 | -.1731643 .8890065 -0.19 0.846 -1.91615 1.569821
10 | .320897 .9012295 0.36 0.722 -1.446053 2.087847
11 | -.4672761 1.125556 -0.42 0.678 -2.674041 1.739489
12 | -.4847717 1.063409 -0.46 0.649 -2.56969 1.600147
13 | -.1433748 1.020529 -0.14 0.888 -2.144222 1.857473
14 | -.2006263 .8903235 -0.23 0.822 -1.946194 1.544941
16 | .8763018 .9196161 0.95 0.341 -.9266968 2.6793
18 | .4877597 .7734222 0.63 0.528 -1.028611 2.004131
21 | .2518529 .8895093 0.28 0.777 -1.492118 1.995824
22 | -.0626013 .9097828 -0.07 0.945 -1.846321 1.721118
23 | .6991119 .9062408 0.77 0.440 -1.077663 2.475887
24 | .8943916 1.196247 0.75 0.455 -1.450969 3.239753
25 | .0302612 .8446954 0.04 0.971 -1.625848 1.68637
26 | -1.074994 1.202115 -0.89 0.371 -3.43186 1.281873
27 | -.1815104 .9229585 -0.20 0.844 -1.991062 1.628041
28 | -.1836055 .9180045 -0.20 0.841 -1.983444 1.616233
30 | .0893199 .8873337 0.10 0.920 -1.650386 1.829026
31 | -.029833 .9358186 -0.03 0.975 -1.864598 1.804932
32 | .985862 1.285665 0.77 0.443 -1.534812 3.506536
33 | .2687769 .8859324 0.30 0.762 -1.468181 2.005735
|
_cons | 2.03513 1.732971 1.17 0.240 -1.362531 5.432792
----------------------------------------------+----------------------------------------------------------------
sigma_u | 3.5765006
sigma_e | 2.8954032
rho | .60408633 (fraction of variance due to u_i)
---------------------------------------------------------------------------------------------------------------
Code with 1-digi industry code
Code:
. xtreg overtimehours ib5.freqWFH4##i.children_in_hh_dummy07 i.isced_edu i.syear i.maritalstatus age agesq i.emp
> status i.disability_status logindincome workingexperience i.jobchange2 i.regtyp i.sizecompany firmtime firmtim
> esq i.partnerWFH i.industrycode if sex==2 & children_in_hh_dummy816==0 , fe vce(cluster pid)
note: age omitted because of collinearity.
Fixed-effects (within) regression Number of obs = 7,765
Group variable: pid Number of groups = 3,737
R-squared: Obs per group:
Within = 0.0417 min = 1
Between = 0.0258 avg = 2.1
Overall = 0.0264 max = 5
F(41, 3736) = .
corr(u_i, Xb) = -0.3373 Prob > F = .
(Std. err. adjusted for 3,737 clusters in pid)
---------------------------------------------------------------------------------------------------------------
| Robust
overtimehours | Coefficient std. err. t P>|t| [95% conf. interval]
----------------------------------------------+----------------------------------------------------------------
freqWFH4 |
Daily | 2.506616 .7575746 3.31 0.001 1.021316 3.991916
Semi-frequent, at least monthly | 1.31014 .5567582 2.35 0.019 .2185599 2.401719
|
1.children_in_hh_dummy07 | -.8776977 .2584498 -3.40 0.001 -1.384414 -.3709813
|
freqWFH4#children_in_hh_dummy07 |
Daily#1 | -1.745719 1.325147 -1.32 0.188 -4.3438 .8523621
Semi-frequent, at least monthly#1 | -2.328925 .9259446 -2.52 0.012 -4.144331 -.5135185
|
isced_edu |
intermediate edu | .7982281 .3598182 2.22 0.027 .0927689 1.503687
higher edu | 1.202107 .4609335 2.61 0.009 .2984012 2.105813
|
syear |
1999 | .0688035 .3655281 0.19 0.851 -.6478506 .7854577
2002 | .8242518 .3520537 2.34 0.019 .1340155 1.514488
2009 | 1.977714 .7760868 2.55 0.011 .4561186 3.499309
2014 | 2.231432 1.111233 2.01 0.045 .0527493 4.410114
|
maritalstatus |
Married, But Separated | .2074701 .4178295 0.50 0.620 -.6117261 1.026666
Single | .0167293 .232257 0.07 0.943 -.4386335 .4720922
Divorced | .1407972 .377668 0.37 0.709 -.5996583 .8812527
Widowed | -.3756726 .5287056 -0.71 0.477 -1.412252 .6609071
Registered same sex partnership | 1.95335 .2149278 9.09 0.000 1.531963 2.374738
Registered same sex partnership, but separ.. | -1.563473 .5525402 -2.83 0.005 -2.646783 -.4801634
|
age | 0 (omitted)
agesq | -.0012707 .0006822 -1.86 0.063 -.0026082 .0000668
|
empstatus |
Regular Part-Time Employment | -.5092677 .2200708 -2.31 0.021 -.9407383 -.0777971
1.disability_status | -.0305753 .3001044 -0.10 0.919 -.6189598 .5578092
logindincome | .2135416 .0907753 2.35 0.019 .0355675 .3915157
workingexperience | -.0529485 .0252211 -2.10 0.036 -.1023971 -.0035
|
jobchange2 |
Yes, changed job in last year | -.1946817 .2089226 -0.93 0.351 -.6042951 .2149317
|
regtyp |
[2] Rural regions | .1158979 .4190269 0.28 0.782 -.7056459 .9374416
|
sizecompany |
2 | .2628441 .245817 1.07 0.285 -.2191045 .7447927
3 | .2407289 .2683291 0.90 0.370 -.285357 .7668149
4 | .2010899 .2576507 0.78 0.435 -.3040599 .7062396
5 | .2992914 .2626124 1.14 0.254 -.2155862 .8141691
6 | .5893057 .2859272 2.06 0.039 .028717 1.149894
7 | .1081272 .3484248 0.31 0.756 -.5749942 .7912486
Unknown | .2916678 .4128469 0.71 0.480 -.5177594 1.101095
|
firmtime | -.026983 .0257914 -1.05 0.296 -.0775496 .0235836
firmtimesq | .0005946 .0007822 0.76 0.447 -.0009389 .0021282
|
partnerWFH |
Partner Working Remotely | .235914 .2128906 1.11 0.268 -.1814791 .6533071
|
industrycode |
2 | -.6843359 1.497694 -0.46 0.648 -3.620714 2.252042
3 | 1.107829 1.78769 0.62 0.535 -2.397113 4.612772
4 | -.2453813 .8045791 -0.30 0.760 -1.822838 1.332076
5 | -.2813446 .8825517 -0.32 0.750 -2.011675 1.448986
6 | .6179438 .7572475 0.82 0.415 -.8667151 2.102603
7 | .1473572 .8765486 0.17 0.867 -1.571203 1.865918
8 | .2111168 .8223362 0.26 0.797 -1.401155 1.823388
9 | -.0472463 .8081457 -0.06 0.953 -1.631696 1.537203
|
_cons | 1.936961 1.723199 1.12 0.261 -1.441541 5.315463
----------------------------------------------+----------------------------------------------------------------
sigma_u | 3.5625995
sigma_e | 2.891451
rho | .60287627 (fraction of variance due to u_i)
---------------------------------------------------------------------------------------------------------------
Comment