{smcl}
{* 18apr2007}
help for {hi:ftest}
{hline}
{title:Title}
{p2colset 5 14 16 2}{...}
{p2col :{hi: ftest} {hline 2}} compare two nested models using an F-test.{p_end}
{p2colreset}{...}
{title:Syntax}
{p 8 15 2}
{cmd:ftest} {it:modelspec1} [{it:modelspec2}]
{phang}
where {it:modelspec} is
{phang2}
{it:name}{c |}{cmd:.}
{phang}
where {it:name} is the name under which estimation results were saved using
{helpb estimates store:estimates store}, and "{cmd:.}" refers to the last
estimation results, whether or not these were already stored.
{title:Description}
{pstd}
{cmd: ftest} compares two nested models estimated using {helpb regress} and
performs an F-test for the null hypothesis that the constraint implict in
the restricted model holds. For example if a variable is left out of the
restricted model, the implict constraint is that the coefficient for that
variable equals zero. To conduct the test, both the unrestricted and the
restricted models must be fitted using {helpb regress}, and the results
of at least one must be stored using
{helpb estimates store:estimates store}.
{pstd}
{it:modelspec1} and {it:modelspec2} specify the restricted and unrestricted
model in any order. If {it:modelspec2} is not specified, the last
estimation result is used; this is equivalent to specifying {it:modelspec2}
as a period ({cmd:.}).
{title:Remarks}
{phang}
{cmd: ftest} is a convenience command; anything that can be done with
{cmd: ftest} can be done with {helpb test}, and it will produce exactly the
same results. The difference is that with test the constraint needs to be
explicitly specified, while with ftest the constraint is implicit. However,
{cmd: test} will work with all {help vce_option:vce types}, while
{cmd: ftest} will only work with the default vce type.
{phang}
{cmd: ftest} requires that both models are estimated on exactly the same
sample. This can easily be violated when the restricted model leaves a
variable out that contains missing values. The solution is to use the
{help if} qualifier in {cmd:regress} when estimating the restricted model,
see the Examples below.
{title:Examples}
{phang}{cmd: Various ways specifying the models}
{phang}{stata "sysuse auto, clear"}{p_end}
{phang}{stata "reg price mpg foreign"}{p_end}
{phang}{stata "est store a"}{p_end}
{phang}{stata "reg price mpg"}{p_end}
{phang}{stata "est store b"}{p_end}
{phang}{stata "ftest a b"}{p_end}
{phang}{stata "ftest a ."}{p_end}
{phang}{stata "ftest a"}{p_end}
{phang}{cmd: This test can also be performed using }{helpb test}
{phang}{stata "est restore a"}{p_end}
{phang}{stata "test foreign"}{p_end}
{phang}{cmd: this will cause an error because rep78 contains missing values}
{phang}{stata "reg price mpg rep78"}{p_end}
{phang}{stata "est store a"}{p_end}
{phang}{stata "reg price mpg"}{p_end}
{phang}{stata "est store b"}{p_end}
{phang}{stata "ftest a b"}{p_end}
{phang}{cmd: this is the solution}
{phang}{stata "reg price mpg rep78"}{p_end}
{phang}{stata "est store a"}{p_end}
{phang}{stata "reg price mpg if e(sample)"}{p_end}
{phang}{stata "est store b"}{p_end}
{phang}{stata "ftest a b"}{p_end}
{title:Author}
{p 4 4}
Maarten L. Buis{break}
Vrije Universiteit Amsterdam{break}
Department of Social Research Methodology{break}
m.buis@fsw.vu.nl
{p_end}
{title:Also see}
{psee}
Online: {helpb regress} {helpb estimates} {helpb test} {helpb lrtest}
{p_end}