Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • outreg2 does not allow me to export F-statistic

    I encountered a weird problem when trying to export results from regression. The code I used is based on the help file of outreg2:
    Code:
    probit access sp500 nyse $xvars utility_ind i.fyear, vce(cluster id)
    test sp500 = nyse = 0
    local fstat = r(F)
    outreg2 using Table7, excel stats(coef se) keep($xvars) nocons alpha(0.01, 0.05, 0.1) symbol(*,**,***) dec(3) ///
    addstat(Pseudo R-squared, `e(r2_p)', F-statistic, `fstat') adec(3) append​​​​​​​
    I always get a "invalid syntax" error for the "addstat" command. The full output is


    ​​​​​​​
    Code:
    . probit access sp500 nyse $xvars utility_ind i.fyear, vce(cluster id)
    
    Iteration 0: log pseudolikelihood = -31829.721
    Iteration 1: log pseudolikelihood = -17721.457
    Iteration 2: log pseudolikelihood = -16911.605
    Iteration 3: log pseudolikelihood = -16892.798
    Iteration 4: log pseudolikelihood = -16892.791
    Iteration 5: log pseudolikelihood = -16892.791
    
    Probit regression Number of obs = 59,912
    Wald chi2(26) = 2273.63
    Prob > chi2 = 0.0000
    Log pseudolikelihood = -16892.791 Pseudo R2 = 0.4693
    
    (Std. Err. adjusted for 9,541 clusters in id)
    ---------------------------------------------------------------------------------
    | Robust
    access | Coef. Std. Err. z P>|z| [95% Conf. Interval]
    ----------------+----------------------------------------------------------------
    sp500 | .4410144 .0800983 5.51 0.000 .2840246 .5980042
    nyse | .2845788 .0443843 6.41 0.000 .1975871 .3715705
    log_mv_assets | .4867428 .0197082 24.70 0.000 .4481153 .5253702
    log_1plus_age | .068075 .0242086 2.81 0.005 .020627 .1155229
    profit_margin | -.1272283 .0347415 -3.66 0.000 -.1953204 -.0591362
    ppe2bv_assets | .136199 .0820912 1.66 0.097 -.0246968 .2970947
    mkt2book | -.1804552 .019395 -9.30 0.000 -.2184687 -.1424417
    rd2sales | -.078013 .0624374 -1.25 0.211 -.2003881 .0443622
    ad2sales | .0955271 .2052 0.47 0.642 -.3066576 .4977117
    lag_cum_ret_use | .0525204 .0125297 4.19 0.000 .0279627 .0770782
    vol_asset | -1.486623 .1096425 -13.56 0.000 -1.701518 -1.271727
    utility_ind | -.0714785 .1198726 -0.60 0.551 -.3064245 .1634675
    |
    fyear |
    1987 | .060118 .0235873 2.55 0.011 .0138878 .1063483
    1988 | .0006752 .0305946 0.02 0.982 -.0592891 .0606394
    1989 | -.1940478 .0335376 -5.79 0.000 -.2597802 -.1283153
    1990 | -.3608065 .0360635 -10.00 0.000 -.4314897 -.2901233
    1991 | -.3422037 .0381232 -8.98 0.000 -.4169237 -.2674836
    1992 | -.3662483 .0388025 -9.44 0.000 -.4422998 -.2901969
    1993 | -.3046348 .0389579 -7.82 0.000 -.3809909 -.2282786
    1994 | -.3791964 .0394049 -9.62 0.000 -.4564285 -.3019643
    1995 | -.410071 .0412203 -9.95 0.000 -.4908613 -.3292807
    1996 | -.3735554 .0417193 -8.95 0.000 -.4553238 -.291787
    1997 | -.2903574 .0426445 -6.81 0.000 -.373939 -.2067758
    1998 | -.2207306 .0429638 -5.14 0.000 -.3049381 -.1365232
    1999 | -.1390364 .0457177 -3.04 0.002 -.2286415 -.0494313
    2000 | -.1995982 .0462516 -4.32 0.000 -.2902497 -.1089467
    |
    _cons | -3.249555 .1199023 -27.10 0.000 -3.484559 -3.014551
    ---------------------------------------------------------------------------------
    Note: 92 failures and 0 successes completely determined.
    
    . test sp500 = nyse = 0
    
    ( 1) [access]sp500 - [access]nyse = 0
    ( 2) [access]sp500 = 0
    
    chi2( 2) = 71.71
    Prob > chi2 = 0.0000
    
    . local fstat = r(F)
    
    . outreg2 using Table7, excel stats(coef se) keep($xvars) nocons alpha(0.01, 0.05, 0.1) symbol(*,**,***) dec(3) ///
    > addstat(Pseudo R-squared, `e(r2_p)', F-statistic, `fstat') adec(3) append
    invalid syntax
    r(198);
    
    end of do-file
    However, when I delete the F-statistic here, i.e., use just addstat(Pseudo R-squared, `e(r2_p)'), it works perfectly. Why does this happen?

  • #2
    The outreg2 command cannot export an F-statistic because test did not produce an F-statistic, it produced a chi-squared statistic, as the displayed results show. In this case the output of help test is not very helpful, but the full documentation for the test command in the Stata Base Reference Manual PDF accessible from Stata's Help menu tells us

    If the estimation command reports significance levels and confidence intervals using z rather than t statistics, test reports results using the chi-squared rather than the F statistic.
    And indeed, your probit output shows that it uses z rather than t statistics.

    But help test does at least tell us that you can access the chi-squared value in r(chi2).

    Comment


    • #3
      Thanks William, this is super helpful!!

      It's confusing that a published paper reported F test statistics in a table of Probit regression.

      Comment

      Working...
      X