.-
help for ^prais^                                              (STB-25: sts10)
.-

Prais-Winsten regression
------------------------

    ^prais^  [depvar [varlist] [^in^ range]] [^, l^evel^(^#^) nol^og ^i^terate^(^#^) t^ol^(^#^)^]


^prais^ share the features of estimation commands (see help @est@),
but also see warning below.  See help @matsize@ to reset problem-size limits.


Description
-----------

^prais^ estimates a linear regression of depvar on varlist that is corrected 
for serially correlated residuals using the Prais-Winsten (1954) iterative
process. This is similar to the Cochrane-Orcutt procedure but maintains the 
first observation and iterates to find the estimate of the correlation.

Warning:  To use ^predict^ after ^prais^, you must first "^gen _iter=1^".


Options
-------

^level(^#^)^ specifies the significance level in percent for confidence intervals.

^nolog^ suppresses the iteration log.

^iterate(^#^)^ specifies the maximum number of iterations and defaults to 100.  You
    should never have to specify this option.

^tol(^#^)^ specifies the minimum change in the estimated autocorrelation parameter
    between iterations before convergence can be declared and defaults
    to .001.


Examples
--------

 . ^regdw sur idel time^      (estimate model, report Durbin-Watson statistic)
 . ^prais sur idel time^      (reestimate model using Prais-Winsten correction)
 . ^prais^                    (redisplay estimates)


Author
------

        James W. Hardin
        Stata Corporation
        702 University Drive East
        Phone: 409-696-4600
               800-782-8272
        Fax:   409-696-4601
        email: stata@@stata.com


Also See
--------

    STB:  STB-25  sts10
 Manual:  [5s] corc
On-line:  help for @correlate@, @predict@, @regress@, @test@