UCLA Academic Technology Services HomeServicesClassesContactJobs
Search

Stata Textbook Examples
Regression Analysis by Example, Third Edition
Chapter 12: Logistic Regression

Part of table 12.1, page 323.  The variable index corresponds to the row value in table 12.1.
use http://www.ats.ucla.edu/stat/stata/examples/chp/p323, clear

generate index = _n
list in f/5

            y         x1         x2         x3      index 
  1.        0      -62.8      -89.5        1.7          1  
  2.        0        3.3       -3.5        1.1          2  
  3.        0     -120.8     -103.2        2.5          3  
  4.        0      -18.1      -28.8        1.1          4  
  5.        0       -3.8      -50.6         .9          5 

list in -5/l

            y         x1         x2         x3      index 
 62.        1       53.1        7.1        1.9         62  
 63.        1       39.8       13.8        1.2         63  
 64.        1       59.5          7          2         64  
 65.        1       16.3       20.4          1         65  
 66.        1       21.7       -7.8        1.6         66  
Page 324, table 12.2.

Note: The first logit command displays the results as coefficients while the second logit command with the or option displays the results as odds ratios.
logit y x1 x2 x3

Logit estimates                                   Number of obs   =         66
                                                  LR chi2(3)      =      85.68
                                                  Prob > chi2     =     0.0000
Log likelihood = -2.9064524                       Pseudo R2       =     0.9365

------------------------------------------------------------------------------
       y |      Coef.   Std. Err.       z     P>|z|       [95% Conf. Interval]
---------+--------------------------------------------------------------------
      x1 |   .3312468   .3007071      1.102   0.271      -.2581284    .9206219
      x2 |   .1808756   .1069197      1.692   0.091      -.0286831    .3904344
      x3 |   5.087464   5.081626      1.001   0.317       -4.87234    15.04727
   _cons |  -10.15345    10.8389     -0.937   0.349       -31.3973     11.0904
------------------------------------------------------------------------------

Note: 17 failures and 10 successes completely determined.

logit, or


Logit estimates                                   Number of obs   =         66
                                                  LR chi2(3)      =      85.68
                                                  Prob > chi2     =     0.0000
Log likelihood = -2.9064524                       Pseudo R2       =     0.9365

------------------------------------------------------------------------------
       y | Odds Ratio   Std. Err.       z     P>|z|       [95% Conf. Interval]
---------+--------------------------------------------------------------------
      x1 |   1.392703   .4187958      1.102   0.271       .7724961    2.510851
      x2 |   1.198266   .1281183      1.692   0.091       .9717243    1.477623
      x3 |   161.9786   823.1147      1.001   0.317       .0076554     3427251
------------------------------------------------------------------------------

Note: 17 failures and 10 successes completely determined.
Logistic regression diagnostics, page 325.
predict p  /* predicted probabilities */

predict pr, rstandard  /* standardized Personian residuals */
(10 missing values generated)

predict dr, deviance  /* deviance residuals */

predict pstar, hat  /* leverage */

predict dbeta, dbeta  /* DBETA */
(10 missing values generated)

predict dg, dx2  /*  change in chi-square when observation deleted */
(10 missing values generated)
Figures 12.2, 12.3 and 12.4, pages 326 and 327.

Note: Observations 9 and 52 appear to be problematic; however, observation 36 does not appear problematic, differing from the figures in the book.
sort index
graph twoway (scatter dr index) (scatter dr index if abs(dr) >= 1, mlabel(index)), ///
		ylabel(-1.5(.75).75) xlabel(15(15)60)
graph twoway (scatter dbeta index) (scatter dbeta index if abs(dbeta) >= 2, mlabel(index)), ///
		xlabel(15(15)60)
graph twoway (scatter dg index) (scatter dg index if abs(dg) >= 2, mlabel(index)), ///
		xlabel(15(15)60)
Likelihood ratio test, page 327 and table 12.3, page 328.
lrtest, saving (0)

logit y x1 x2

Logit estimates                                   Number of obs   =         66
                                                  LR chi2(2)      =      82.02
                                                  Prob > chi2     =     0.0000
Log likelihood = -4.7359473                       Pseudo R2       =     0.8965

------------------------------------------------------------------------------
       y |      Coef.   Std. Err.       z     P>|z|       [95% Conf. Interval]
---------+--------------------------------------------------------------------
      x1 |   .1573639   .0749183      2.100   0.036       .0105267    .3042011
      x2 |   .1947428   .1224281      1.591   0.112      -.0452119    .4346974
   _cons |  -.5503398   .9509818     -0.579   0.563       -2.41423     1.31355
------------------------------------------------------------------------------

Note: 11 failures and 0 successes completely determined.

lrtest

Logit:  likelihood-ratio test                         chi2(1)     =       3.66
                                                      Prob > chi2 =     0.0558
Likelihood ratio test, page 328 and table 12.4, page 328.
lrtest, saving(1)

logit y x1

Logit estimates                                   Number of obs   =         66
                                                  LR chi2(1)      =      75.69
                                                  Prob > chi2     =     0.0000
Log likelihood = -7.9015445                       Pseudo R2       =     0.8273

------------------------------------------------------------------------------
       y |      Coef.   Std. Err.       z     P>|z|       [95% Conf. Interval]
---------+--------------------------------------------------------------------
      x1 |   .1767191   .0571025      3.095   0.002       .0648004    .2886379
   _cons |  -1.166587   .8163989     -1.429   0.153        -2.7667    .4335252
------------------------------------------------------------------------------

Note: 9 failures and 0 successes completely determined.

lrtest, using(1)

Logit:  likelihood-ratio test                         chi2(1)     =       6.33
                                                      Prob > chi2 =     0.0119
Table 12.5, page 331.
regress y x1 x2 x3

  Source |       SS       df       MS                  Number of obs =      66
---------+------------------------------               F(  3,    62) =   27.38
   Model |  9.40333712     3  3.13444571               Prob > F      =  0.0000
Residual |  7.09666288    62  .114462305               R-squared     =  0.5699
---------+------------------------------               Adj R-squared =  0.5491
   Total |       16.50    65  .253846154               Root MSE      =  .33832

------------------------------------------------------------------------------
       y |      Coef.   Std. Err.       t     P>|t|       [95% Conf. Interval]
---------+--------------------------------------------------------------------
      x1 |   .0031216   .0008294      3.764   0.000       .0014637    .0047795
      x2 |   .0042457   .0014367      2.955   0.004       .0013737    .0071177
      x3 |   .1485037   .0453154      3.277   0.002       .0579195    .2390878
   _cons |   .3218658   .0874576      3.680   0.000       .1470407     .496691
------------------------------------------------------------------------------
Table 12.6, page 332.

Note: the command generate a = p >= .5 creates a new variable a which has the value 1 if p is greater or equal to .5 and 0 otherwise.
predict p
generate a = p >= .5
list y p a

            y          p          a  
  1.        0  -.0017021          0  
  2.        0   .4806612          0  
  3.        0   -.122117          0  
  4.        0   .3064435          0  
  5.        0   .2288255          0  
  6.        0   .1446737          0  
  7.        0   .3331264          0  
  8.        0   -.320569          0  
  9.        0    .517042          1  
 10.        0   .1161951          0  
 11.        0   .2255089          0  
 12.        0  -.0724799          0  
 13.        0  -.8029544          0  
 14.        0    .545396          1  
 15.        0   .0301355          0  
 16.        0  -.4522537          0  
 17.        0   .6364052          1  
 18.        0   .4545762          0  
 19.        0    .437882          0  
 20.        0   .1398092          0  
 21.        0   .2249177          0  
 22.        0   .3714215          0  
 23.        0   .1800966          0  
 24.        0   .0519476          0  
 25.        0   .5541553          1  
 26.        0   .5563999          1  
 27.        0    .394581          0  
 28.        0   .3399023          0  
 29.        0   .3882431          0  
 30.        0    .262901          0  
 31.        0   .3894134          0  
 32.        0    .117767          0  
 33.        0   .4403123          0  
 34.        1   .7187779          1  
 35.        1   .8186681          1  
 36.        1   .7295072          1  
 37.        1   .8015884          1  
 38.        1   .6547927          1  
 39.        1   .7962746          1  
 40.        1   .7478317          1  
 41.        1   .7593519          1  
 42.        1   .8322027          1  
 43.        1   1.099869          1  
 44.        1   1.420885          1  
 45.        1   .8599006          1  
 46.        1    .657912          1  
 47.        1   .8058342          1  
 48.        1   .5835429          1  
 49.        1   .9690604          1  
 50.        1   1.029698          1  
 51.        1   .7686977          1  
 52.        1   .4763457          0  
 53.        1   .5957797          1  
 54.        1   .7405336          1  
 55.        1   .8092653          1  
 56.        1   .8389035          1  
 57.        1   .6227317          1  
 58.        1   .8070596          1  
 59.        1   .7396857          1  
 60.        1    .843024          1  
 61.        1   .8565065          1  
 62.        1   .7999232          1  
 63.        1   .6828997          1  
 64.        1   .8343272          1  
 65.        1   .6078632          1  
 66.        1   .5940937          1  

How to cite this page

Report an error on this page

UCLA Researchers are invited to our Statistical Consulting Services
We recommend others to our list of Other Resources for Statistical Computing Help
These pages are Copyrighted (c) by UCLA Academic Technology Services


The content of this web site should not be construed as an endorsement of any particular web site, book, or software product by the University of California