vwls.hlp

是一个经济学管理应用软件 很难找的 但是经济学学生又必须用到

{smcl}
{* 17mar2005}{...}
{cmd:help vwls}{right:dialog:  {bf:{dialog vwls}}{space 15}}
{right:also see:  {help vwls postestimation}}
{hline}
 
{title:Title}

{p2colset 5 17 19 2}{...}
{p2col :{hi:[R] vwls} {hline 2}}Variance-weighted least squares{p_end}
{p2colreset}{...}


{title:Syntax}

{p 8 16 2}
{cmd:vwls} {depvar} {indepvars} {ifin} {weight} 
[{cmd:,} {it:options}]

{synoptset 14 tabbed}{...}
{synopthdr:options}
{synoptline}
{syntab:Model}
{synopt :{opt nocon:stant}}suppress constant term{p_end}
{synopt :{opth sd(varname)}}variable containing estimate of conditional
standard deviation{p_end}

{syntab:Reporting}
{synopt :{opt l:evel(#)}}set confidence level; default is {cmd:level(95)}{p_end}
{synoptline}
{p2colreset}{...}
{p 4 6 2}{cmd:bootstrap}, {cmd:by}, {cmd:jackknife}, {cmd:rolling}, 
{cmd:statsby}, and {cmd:xi} are allowed; see {help prefix}.{p_end}
{p 4 6 2}{cmd:fweight}s are allowed; see {help weight}.{p_end}
{p 4 6 2}See {help vwls postestimation} for additional capabilities of 
estimation commands.{p_end}


{title:Description}

{pstd}
{cmd:vwls} estimates a linear regression using variance-weighted least
squares.  It differs from ordinary least-squares (OLS) regression in that it
does not assume homogeneity of variance, but requires that the conditional
variance of {depvar} be estimated prior to the regression.  The estimated
variance need not be constant across observations.  {cmd:vwls} treats the
estimated variance as if it were the true variance when it computes standard
errors of the coefficients.

{pstd}
You must supply an estimate of the conditional standard deviation of {depvar}
to {cmd:vwls} using the {opth sd(varname)} option, or you must have grouped
data with groups defined by the {indepvars} variables.  In the latter case,
{cmd:vwls} treats all {it:indepvars} as categorical variables, computes the
mean and standard deviation of {it:depvar} separately for each subgroup; and 
computes the regression of the subgroup means on {it:indepvars}. 

{pstd}
The first consists of measurements from physical science experiments in
which all error is due solely to measurement error and the sizes of the
measurement errors are known.

{pstd}
The second concerns certain categorical data analyses.  The independent
variables are categorical and the outcome can be sensibly averaged.  Assuming
each of the subgroups defined by the independent variables contains a
reasonable number of subjects, then the variance of the outcome variable can
be estimated independently within each group.

{pstd}
{cmd:regress} with analytic weights can be used to produce another kind of
"variance-weighted least squares"; see the following remarks for an
explanation of the difference. 


{title:Options}

{dlgtab:Model}

{phang}
{opt noconstant}; see {help estimation options##noconstant:estimation options}. 

{phang}
{opth sd(varname)} specifies an estimate of the conditional standard deviation
of {depvar} (that is, it can vary observation by observation).  All values of
{it:varname} must be > 0.  If you specify {opt sd()}, you cannot use
{cmd:fweight}s.

{pmore}
If {opt sd()} is not given, the data will be grouped by {indepvars}.  In this
case, {it:indepvars} are treated as categorical variables, and the means and
standard deviations of {it:depvars} for each subgroup are calculated and used
for the regression.  Any subgroup for which the standard deviation is zero is
dropped.

{dlgtab:Reporting}

{phang}
{opt level(#)}; see {help estimation options##level():estimation options}. 


{title:Example}

    {bf:Example 1}

{pmore}
An experimenter measures the quantities {cmd:x} and {cmd:y} and estimates that
the standard deviation of each {cmd:y} measurement is {cmd:stderr}:

                {cmd:x        y   stderr}
                {hline 19}
                1       1.2     0.5
                3       3.2     1.0
                4       4.3     1.0
                ...

{phang3}{cmd:. vwls y x, sd(stderr)}


    {bf:Example 2}

{phang3}. vwls bp gender race

{pmore}
where {cmd:gender} and {cmd:race} are categorical (0/1) variables.  If
{cmd:race} were a categorical variable taking on three values,

{phang3}{cmd:. xi: vwls bp gender i.race}

{pmore}
See {helpb xi}.


{title:Also see}

{psee}
Manual:  {bf:[R] vwls}

{psee}
Online:  {help vwls postestimation};{break}
{helpb regress}{p_end}