Predictions from generalized linear model                     [STB-14: sg16.2]
-----------------------------------------

^gpredict^ newvarname, [^m^|^e^|^p^|^dev^|^sp^|^sd^|^h^|^l^|^an^|^dp^|^dd^|^db^]

where options and appropriate distributions are:

^Mu^       = predicted fit         [all]
^Eta^      = linear predictor      [all]
^Pearson^  = Pearson residual      [all]
^DEVianc^  = deviance residual     [all]
^SPear^    = standardized Pearson  [all]
^SDev^     = standardized deviance [all]
^Hat^      = hat matrix diagonal   [all]   
^ANscomb^  = Anscombe residual     [bin,poi,gam,ivg]
^Likelih^  = likelihood residual   [bin] 
^DPears^   = delta Pearson         [bin]
^DDevian^  = delta deviance        [bin]
^DBeta^    = delta beta            [bin]



^gpredict^ is to be used following the ^glm^ command.  For example,

     ^glm kyph age start, f(bin) l(l)
     ^gpredict m,m 
     ^gpredict dp,dp
     ^graph dp m, xlab ylab yline(0)


Author
------

Joseph Hilbe              email: atjmh@@asuvm.inre.asu.edu
Dept of Sociology         fax:   602-860-1446
Arizona State University
Tempe, AZ 85287


References
----------

Collett, D., 1991.  ^Modelling binary data^, New York:Chapman & Hall.

Hilbe, J., 1993. ^sg16: Generalized linear models^, Stata Technical Bulletin,
        (STB-11: Jan 1993).

McCullagh and Nelder, 1989.  ^Generalized linear models^, New York:
        Chapman & Hall.