gretl version 1.9.12
Current session: 2013-04-09 16:50
# ESERCIZI E4.1, E5.1, E6.1
# OCCHIO, questo percorso vale sul mio computer, sul vostro sara' diverso
? open "/Users/fausto/Library/Application \
  Support/gretl/data/stock_watson_2/CPS04.gdt"

Read datafile /Users/fausto/Library/Application Support/gretl/data/stock_watson_2/CPS04.gdt
periodicity: 1, maxobs: 7986
observations range: 1 to 7986

Listing 5 variables:
  0) const       1) ahe         2) bachelor    3) female      4) age       

# model 1
? ols ahe const age --robust

Model 1: OLS, using observations 1-7986
Dependent variable: ahe
Heteroskedasticity-robust standard errors, variant HC1

             coefficient   std. error   t-ratio   p-value 
  --------------------------------------------------------
  const       3.32418      0.965696      3.442    0.0006   ***
  age         0.451931     0.0329673    13.71     2.72e-42 ***

Mean dependent var   16.77115   S.D. dependent var   8.758696
Sum squared resid    598935.5   S.E. of regression   8.661234
R-squared            0.022254   Adjusted R-squared   0.022131
F(1, 7984)           187.9219   P-value(F)           2.72e-42
Log-likelihood      −28571.28   Akaike criterion     57146.55
Schwarz criterion    57160.52   Hannan-Quinn         57151.33

# Bob
? genr yhat_bob = $coeff(const) + $coeff(age)*26
Generated scalar yhat_bob = 15.0744
# Alexis
? genr yhat_alexis = $coeff(const) + $coeff(age)*30
Generated scalar yhat_alexis = 16.8821
# No solo il 2% della varianza totale e' spiegato da age
# Si', lo slope e' significativo
# intervallo di confidenza per lo slope al 95%
? genr CIslope_hi = $coeff(age)+1.96*$stderr(age)
Generated scalar CIslope_hi = 0.516547
? genr CIslope_lo = $coeff(age)-1.96*$stderr(age)
Generated scalar CIslope_lo = 0.387315
# Campione ristretto ai diplomati
? smpl bachelor==0 --restrict
Full data set: 7986 observations
Current sample: 4346 observations
# model 2
? ols ahe const age --robust

Model 2: OLS, using observations 1-4346
Dependent variable: ahe
Heteroskedasticity-robust standard errors, variant HC1

             coefficient   std. error   t-ratio   p-value 
  --------------------------------------------------------
  const       6.20009      1.01564       6.105    1.12e-09 ***
  age         0.255720     0.0344032     7.433    1.27e-13 ***

Mean dependent var   13.80961   S.D. dependent var   6.729093
Sum squared resid    194362.1   S.E. of regression   6.688996
R-squared            0.012109   Adjusted R-squared   0.011882
F(1, 4344)           55.25011   P-value(F)           1.27e-13
Log-likelihood      −14425.12   Akaike criterion     28854.24
Schwarz criterion    28867.00   Hannan-Quinn         28858.75

# Back to full sample
? smpl full
Full data range: 1 - 7986 (n = 7986)

# Restringiamo il campione ai laurati
? smpl bachelor==1 --restrict
Full data set: 7986 observations
Current sample: 3640 observations
# model 3
? ols ahe const age --robust

Model 3: OLS, using observations 1-3640
Dependent variable: ahe
Heteroskedasticity-robust standard errors, variant HC1

             coefficient   std. error   t-ratio   p-value 
  --------------------------------------------------------
  const       −0.232686    1.54781      −0.1503   0.8805  
  age          0.690387    0.0528546    13.06     3.86e-38 ***

Mean dependent var   20.30709   S.D. dependent var   9.554442
Sum squared resid    317747.7   S.E. of regression   9.345658
R-squared            0.043490   Adjusted R-squared   0.043227
F(1, 3638)           170.6158   P-value(F)           3.86e-38
Log-likelihood      −13299.01   Akaike criterion     26602.03
Schwarz criterion    26614.43   Hannan-Quinn         26606.45

# Back to full sample
? smpl full
Full data range: 1 - 7986 (n = 7986)

# Si', per i laureati l'effetto eta' e' maggiore
# model 4
? ols ahe const age female bachelor --robust

Model 4: OLS, using observations 1-7986
Dependent variable: ahe
Heteroskedasticity-robust standard errors, variant HC1

             coefficient   std. error   t-ratio    p-value 
  ---------------------------------------------------------
  const        1.88380     0.897242       2.100   0.0358    **
  age          0.439204    0.0301511     14.57    1.85e-47  ***
  female      −3.15786     0.175588     −17.98    6.39e-71  ***
  bachelor     6.86515     0.185029      37.10    3.84e-278 ***

Mean dependent var   16.77115   S.D. dependent var   8.758696
Sum squared resid    496180.7   S.E. of regression   7.884317
R-squared            0.189998   Adjusted R-squared   0.189694
F(3, 7982)           545.3021   P-value(F)           0.000000
Log-likelihood      −27819.74   Akaike criterion     55647.48
Schwarz criterion    55675.42   Hannan-Quinn         55657.04

# IC per il coeff di age
? genr CIslope_hi = $coeff(age)+1.96*$stderr(age)
Replaced scalar CIslope_hi = 0.4983
? genr CIslope_lo = $coeff(age)-1.96*$stderr(age)
Replaced scalar CIslope_lo = 0.380108
# Bob ancora
? genr yhat_bob_again = $coeff(const) + $coeff(age)*25 + $coeff(female)*0 + \
  $coeff(bachelor)*0
Generated scalar yhat_bob_again = 12.8639
# Alexis ancora
? genr yhat_alexis_again = $coeff(const) + $coeff(age)*30 + $coeff(female)*1 + \
  $coeff(bachelor)*1
Generated scalar yhat_alexis_again = 18.7672
# F test
? omit female bachelor --test-only
Test on Model 4:

  Null hypothesis: the regression parameters are zero for the variables
    female, bachelor
  Test statistic: Robust F(2, 7982) = 752.951, p-value 2.76557e-300