I have the following stylized example.
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input float(student_performance student_female evaluator_female dyad_stud_eval) long(evaluator_num student_num) 12 1 1 2 2 1 13 1 1 2 2 1 14 1 1 4 2 2 10 1 1 4 2 2 8 1 0 3 1 2 9 0 0 5 1 3 2 0 1 6 2 3 12 0 0 7 1 4 9 0 0 7 1 4 9 1 0 1 1 1 10 1 1 4 2 2 11 1 1 2 2 1 end label values evaluator_num evaluator_num label def evaluator_num 1 "X", modify label def evaluator_num 2 "Y", modify label values student_num student_num label def student_num 1 "A", modify label def student_num 2 "B", modify label def student_num 3 "C", modify label def student_num 4 "D", modify
I want to interact the gender of the student with the gender of the evaluator and see if there is any moderation going on that effects the students performance.
Now I am thinking about the fixed-effect level and I would like to run both, student and evaluator fe since evaluators might differ in students evaluation besides gender (same for students).
Now I have several options to run the regressions:
Code:
reg student_performance student_female##evaluator_female
Code:
Source | SS df MS Number of obs = 12
-------------+---------------------------------- F(3, 8) = 11.44
Model | 85.0833333 3 28.3611111 Prob > F = 0.0029
Residual | 19.8333333 8 2.47916667 R-squared = 0.8110
-------------+---------------------------------- Adj R-squared = 0.7401
Total | 104.916667 11 9.53787879 Root MSE = 1.5745
-------------------------------------------------------------------------------------------------
student_performance | Coef. Std. Err. t P>|t| [95% Conf. Interval]
--------------------------------+----------------------------------------------------------------
1.student_female | -1.5 1.437349 -1.04 0.327 -4.814533 1.814533
1.evaluator_female | -8 1.818119 -4.40 0.002 -12.19259 -3.807411
|
student_female#evaluator_female |
1 1 | 11.16667 2.226732 5.01 0.001 6.031815 16.30152
|
_cons | 10 .9090593 11.00 0.000 7.903705 12.09629
-------------------------------------------------------------------------------------------------
Code:
areg student_performance student_female##evaluator_female, absorb(dyad_stud_eval)
Code:
Linear regression, absorbing indicators Number of obs = 12
Absorbed variable: dyad_stud_eval No. of categories = 7
F( 0, 5) = .
Prob > F = .
R-squared = 0.8364
Adj R-squared = 0.6400
Root MSE = 1.8529
-------------------------------------------------------------------------------------------------
student_performance | Coef. Std. Err. t P>|t| [95% Conf. Interval]
--------------------------------+----------------------------------------------------------------
1.student_female | 0 (omitted)
1.evaluator_female | 0 (omitted)
|
student_female#evaluator_female |
1 1 | 0 (omitted)
|
_cons | 9.916667 .5348936 18.54 0.000 8.541679 11.29165
-------------------------------------------------------------------------------------------------
F test of absorbed indicators: F(6, 5) = 4.260 Prob > F = 0.067
Code:
reg student_performance student_female##evaluator_female i.student_num i.evaluator_num
Code:
Source | SS df MS Number of obs = 12
-------------+---------------------------------- F(5, 6) = 6.12
Model | 87.7083333 5 17.5416667 Prob > F = 0.0238
Residual | 17.2083333 6 2.86805556 R-squared = 0.8360
-------------+---------------------------------- Adj R-squared = 0.6993
Total | 104.916667 11 9.53787879 Root MSE = 1.6935
-------------------------------------------------------------------------------------------------
student_performance | Coef. Std. Err. t P>|t| [95% Conf. Interval]
--------------------------------+----------------------------------------------------------------
1.student_female | -1.625 1.796263 -0.90 0.401 -6.020298 2.770298
1.evaluator_female | -7 2.395018 -2.92 0.027 -12.8604 -1.139602
|
student_female#evaluator_female |
1 1 | 10.16667 2.765529 3.68 0.010 3.399662 16.93367
|
student_num |
B | -.75 1.197509 -0.63 0.554 -3.680199 2.180199
C | -1.5 2.074146 -0.72 0.497 -6.575253 3.575253
D | 0 (omitted)
|
evaluator_num |
Y | 0 (omitted)
_cons | 10.5 1.197509 8.77 0.000 7.569801 13.4302
-------------------------------------------------------------------------------------------------
Sorry for the long text!
Bests,
Julian
0 Response to Fixed-effect level: Is student AND evaluator fixed-effect still correct?
Post a Comment