{smcl}
{* 30 december 2005}{...}
{hline}
help for {hi:hlm, mkmdm, mdmset, hlm2, hlm3, run}{right:sean f. reardon}
{hline}
{title:Writing and Running HLM v.6 files from within Stata}
{p 4 15 2}{cmd:hlm mkmdm} using [{cmd:if} {it:exp}] [{cmd:in} {it:range}]
{cmd:,}
{cmdab:id2(}{it:varname}{cmd:)}
[{cmdab:t:ype(}{it:string}{cmd:)}
{cmd:id3(}{it:varname}{cmd:)}
{cmd:l1(}{it:varlist}{cmd:)}
{cmd:l2(}{it:varlist}{cmd:)}
{cmd:l3(}{it:varlist}{cmd:)}
{cmdab:m:iss(now }{it:or}{cmd: analysis)}
{cmdab:nodta}
{cmdab:nomdmt}
{cmdab:nosts}
{cmd:replace}
{cmdab:norun}]
{p 4 15 2}{cmd:hlm mdmset} {it:existing .mdm filename}
{p 4 15 2}{cmd:hlm mdmset} {cmd:_drop}
{p 4 15 2}{cmd:hlm hlm2} {it:depvar} {it: model}{cmd:,}
{cmdab:cmd:file:(}{it:newfile}{cmd:)}
[{cmdab:mdm:file:(}{it:existing .mdm filename}{cmd:)}
{cmdab:out:file:(}{it:newfile}{cmd:)}
{cmd:run}
{cmd:replace}
{cmdab:lin:ear}
{cmdab:ber:nouli}
{cmdab:poi:sson}
{cmdab:bin:omial}
{cmdab:mul:tinomial}
{cmdab:ord:inal}
{cmdab:nc:at:(}{it:integer}{cmd:)}
{cmdab:ev:ar:(}{it:varname}{cmd:)}
{cmd:fml}
{cmdab:pr:ompt}
{cmdab:st:op}
{cmd:od}
{cmdab:lap:lace}
{cmd:l6}
{cmdab:ln:um:(}{it:integer}{cmd:)}
{cmdab:maci:t:(}{it:integer}{cmd:)}
{cmdab:mici:t:(}{it:integer}{cmd:)}
{cmdab:macs:top:(}{it:#}{cmd:)}
{cmdab:mics:top:(}{it:#}{cmd:)}
{cmd:wgt1(}{it:varname}{cmd:)}
{cmd:wgt2(}{it:varname}{cmd:)}
{cmd:pwgt(}{it:varname}{cmd:)}
{cmd:fsig(}{it:real}{cmd:)}
{cmdab:te:st:(}{it:test_specification}{cmd:)}
{cmdab:dev:iance:(}{it:real integer}{cmd:)}
{cmdab:st:ore:(lin|us|pa|eml|l6)}
{cmd:robust}
{cmd:noeq}
{cmdab:lo:ng}
{cmd:res1}
{cmd:rv1(}{it:varlist}{cmd:)}
{cmd:res2}
{cmd:rv2(}{it:varlist}{cmd:)}
{cmdab:ti:tle:(}{it:string}{cmd:)}
{cmdab:gr:aph:(}{it:newfile}{cmd:)}]
{p 4 15 2}{cmd:hlm hlm3} {it:depvar} {it: model}{cmd:,}
{cmdab:cmd:file:(}{it:newfile}{cmd:)}
[{cmdab:mdm:file:(}{it:existing .mdm filename}{cmd:)}
{cmdab:out:file:(}{it:newfile}{cmd:)}
{cmd:run}
{cmd:replace}
{cmdab:lin:ear}
{cmdab:ber:nouli}
{cmdab:poi:sson}
{cmdab:bin:omial}
{cmdab:mul:tinomial}
{cmdab:ord:inal}
{cmdab:nc:at:(}{it:integer}{cmd:)}
{cmdab:ev:ar:(}{it:varname}{cmd:)}
{cmd:fml}
{cmdab:pr:ompt}
{cmdab:st:op}
{cmd:od}
{cmdab:lap:lace}
{cmdab:ln:um:(}{it:integer}{cmd:)}
{cmdab:maci:t:(}{it:integer}{cmd:)}
{cmdab:mici:t:(}{it:integer}{cmd:)}
{cmdab:macs:top:(}{it:#}{cmd:)}
{cmdab:mics:top:(}{it:#}{cmd:)}
{cmd:wgt1(}{it:varname}{cmd:)}
{cmd:wgt2(}{it:varname}{cmd:)}
{cmd:wgt3(}{it:varname}{cmd:)}
{cmd:pwgt(}{it:varname}{cmd:)}
{cmd:fsig(}{it:real}{cmd:)}
{cmdab:te:st:(}{it:test_specification}{cmd:)}
{cmdab:dev:iance:(}{it:real integer}{cmd:)}
{cmdab:st:ore:(lin|us|pa|eml)}
{cmd:robust}
{cmd:noeq}
{cmdab:lo:ng}
{cmd:res1}
{cmd:rv1(}{it:varlist}{cmd:)}
{cmd:res2}
{cmd:rv2(}{it:varlist}{cmd:)}
{cmd:res3}
{cmd:rv3(}{it:varlist}{cmd:)}
{cmdab:ti:tle:(}{it:string}{cmd:)}
{cmdab:gr:aph:(}{it:newfile}{cmd:)}]
{p 4 15 2}{cmd:hlm run} {it:existing .hlm filename}
[, {cmd:hlm2}
{cmd:hlm3}
{cmd:mdm(}{it:mdmfile}{cmd:)}
{cmdab:st:ore:(lin|us|pa|eml|l6)}
{cmd:robust}
{cmd:noview}]
In the {cmd:hlm2} command, {it:model} is an expression of the form
{p 2 0 2}[int[({it:L2equation})]] {it:L1var}[({it:L2equation})] [{it:L1var}[({it:L2equation})] ...]] [rand]
where {it:L2equation} is an expression of the form
{p 2 0 2}[int] [{it:L2 varlist}] [rand]
{p 0 0 2}If {it:L2equation} is not specified, it is assumed to be {bf:(int)}.
Any level-1 variable may be group-mean centered by preceding the variable name with an {it:#}.
Any level-1 or level-2 variable may be grand-mean centered by preceding the variable name with an {it:%}.
In the {cmd:hlm3} command, {it:model} is an expression of the form
{p 2 0 2}[int[({it:L2equation})]] [{it:L1var}[({it:L2equation})] [{it:L1var}[({it:L2equation})] ...]] [rand]
where {it:L2equation} is an expression of the form
{p 2 0 2}[int[({it:L3equation})]] [{it:L2var}[({it:L3equation})] [{it:L2var}[({it:L3equation})] ...]] [rand]
If {it:L2equation} is not specified, it is assumed to be {bf:(int(int))}
where {it:L3equation} is an expression of the form
{p 2 0 2}[int] [{it:L3 varlist}] [rand]
{p 0 0 2}If {it:L3equation} is not specified, it is assumed to be {bf:(int)}.
Any level-1 or level-2 variable may be group-mean centered by preceding the variable name with an {it:#}.
Any level-1, level-2, or level-3 variable may be grand-mean centered by preceding the variable name with an {it:%}.
{p 0 0 2}See the {bf:Examples} section below for examples of the {it:model} expression.
{p 0 0 2}{bf:Note}: HLM requires that the level-1 variables in the model be listed in the order that they
appear in the .mdm file. If {bf:int} is included in the model, it must be listed first; likewise, if
{bf:rand} is included in the model, it must be listed last ({bf:rand} can be excluded from the level-1 model,
since HLM automatically assumes it is there). Likewise, within each level-2 and level-3 equation,
the variables must be listed in the order they appear in the .mdm file.
{title:Description}
{p 0 0 2}These commands allow the user to invoke and run HLM v6 from within Stata 8.2 or higher.
In order for this program to run properly, the HLM directory must be included in the PATH.
To see how to include HLM in the PATH variable, see
{browse "http://www.ats.ucla.edu/stat/hlm/faq/path.htm":this page} on how to add
HLM to the PATH variable.
{p 0 0 2}{cmd:hlm mkmdm} creates an MDM file for HLM. HLM uses an MDM file to perform analysis of univariate
hierarchical models and to perform analysis of multivariate hierarchical models. By default,
an MDM file for a 2-level univariate model is created. You can specify the type of MDM file
by using the {cmd:type} option. After an MDM file is created, use {cmd:hlm mdmset} to set the .mdm file
in memory for use in writing and running HLM command files. Use {cmd:hlm mdmset _drop} to drop the current
.mdm file from memory. The commands {cmd:hlm hlm2} and {cmd:hlm hlm3}
are used to write and run HLM command (.hlm) files. Finally, the command {cmd:hlm run} will
run an existing HLM command file.
{p 0 0 2}{cmd:hlm hlm2}, {cmd:hlm hlm3}, and {cmd:hlm run} are e-class programs. Saved
results can be obtained after estimation by use of the {help ereturn} commands.
{title:Options}
{marker par}
{bf:{dlgtab:options for hlm mkmdm}}
{p 0 4}{cmd:id2(}{it:varname}{cmd:)} is required. {it:Varname} specifies the level 2 id.
{p 0 4}{cmd:type(}{it:string}{cmd:)} specifies the type of HLM structure. The possible types are HLM2, HLM3,
HMLM2 and HMLM3. The default is HLM2. {bf:Note}: Currently only the HLM2 and HLM3 options are enabled.
{p 0 4}{cmd:id3(}{it:varname}{cmd:)} specifies the level 3 id variable if the model is a 3-level model.
{p 0 4}{cmd:l1(}{it:varlist}{cmd:)} specifies the level 1 variables to be included in the MDM file. If omitted, all variables
except {it:id2} and {it:id3} are included at level 1.
{p 0 4}{cmd:l2(}{it:varlist}{cmd:)} specifies the level 2 variables to be included in the MDM file. If omitted, all variables
except {it:id2} and {it:id3} are included at level 2.
{p 0 4}{cmd:l3(}{it:varlist}{cmd:)} specifies the level 3 variables to be included in the MDM file. If omitted, all variables
except {it:id2} and {it:id3} are included at level 3.
{p 0 4}{cmd:miss(now}{it: or }{cmd:analysis)} is optional in HLM2 and HLM3 to indicate how to deal with missing data.
{cmd:miss(now)} tells HLM to delete missing data when making the MDM file; {cmd:miss(analysis)} tells HLM to delete
missing data when running analyses. If {cmd:miss} is not specified, HLM assumes there is no missing data.
{p 0 4}{cmd:nodta} specifies that the stata data file used to create the MDM file should be deleted;
the default is to save it.
{p 0 4}{cmd:nomdmt} specifies that the mdmt file should be deleted; the default is to save it.
{p 0 4}{cmd:nosts} specifies that the summary statistics file should be deleted;
the default is to save it.
{p 0 4}{cmd:replace} specifies that the MDM file should be overwritten if it already exists.
{p 0 4}{cmd:norun} specifies that the mdmt file should be created, but not run.
This may be useful for creating MDM files in batch mode or if the mdmt file
needs to be manually changed prior to creating the MDM file. Note that if
{cmd:norun} is specified, the {cmd:nomdmt}, {cmd:nosts}, and {cmd:nodta} options
are ignored (i.e., the .mdmt file is saved, the .dta file is saved, since HLM
needs it to make the .mdm file when the .mdmt file is run later; the {cmd:nosts}
option is irrelevant in this case.)
{marker par}
{bf:{dlgtab:options for hlm mdmset}}
{p 0 4}{cmd:hlm mdmset _drop} instructs Stata to drop the current mdm file name from memory.
{marker par}
{bf:{dlgtab:options for hlm2 and hlm3}}
{p 0 4}{cmdab:cmd:file:(}{it:newfile}{cmd:)} is required. The HLM command file is written to {it:newfile}.hlm.
{p 0 4}{cmd:mdm(}{it:mdmfile}{cmd:)} specifies that {it:mdmfile} is the MDM file
to be used. If not specified, the MDM file specified in the most recent
{cmd:mdmdset} command is used.
{p 0 4}{cmdab:out:file:(}{it:newfile}{cmd:)} specifies that the HLM output is to be written to {it:newfile}.txt.
If {cmdab:out:file:(}{it:newfile}{cmd:)} is not specified, the HLM output is written to {it:filename}.txt,
where {it:filename} is the filename specified in the {cmdab:cmd:file:(}{it:newfile}{cmd:)} command.
{p 0 4}{cmd:run} specifies that the HLM command file is to be run immediately; the default is not to run the
command file.
{p 0 4}{cmd:replace} specifies that the command and output files are to be replaced if they already exist.
{p 0 4}{cmdab:lin:ear} specifies the model is linear. This is the default.
{p 0 4}{cmdab:ber:nouli} specifies the model is bernouli (0 or 1).
{p 0 4}{cmdab:poi:sson} specifies a poisson model (constant or variable exposure).
{p 0 4}{cmdab:bin:omial} specifies a binomial model.
{p 0 4}{cmdab:mul:tinomial} specifies a multinomial model.
{p 0 4}{cmdab:ord:inal} specifies an ordinal model.
{p 0 4}{cmdab:nc:at:(}{it:integer}{cmd:)} specifies the number of categories of the dependent variable in a multinomial
or ordinal model. It must be between 3 and 9.
{p 0 4}{cmdab:ev:ar:(}{it:varname}{cmd:)} specifies the exposure variable in a variable-exposure poisson model or
a binomial (number of trials) model. If {cmd:poisson} is specified and {cmd:evar:(}{it:varname}{cmd:)}
is omitted, a constant-exposure poisson model is estimated; if {cmd:evar:(}{it:varname}{cmd:)} is specified, a
variable-exposure poisson model is estimated.
{p 0 4}{cmd:fml} is only available in {cmd:hlm2}. It specifies that the model is to be
estimated using full maximum likelihood. The default in {cmd:hlm2} is to use restricted
maximum likelihood.
{p 0 4}{cmdab:pr:ompt} specifies that HLM should prompt the user if the maximum number of iterations is reached without
convergence. The default is to continue iterating until convergence, regarless of the number of iterations set in
{cmd:macit}.
{p 0 4}{cmdab:st:op} specifies that HLM should stop iterating if the maximum number of iterations is reached without
convergence. The default is to continue iterating until convergence, regarless of the number of iterations set in
{cmd:macit}.
{p 0 4}{cmd:od} specifies the bernouli or poisson model should be estimated with overdispersion of the level-1 variance.
This option is not allowed with any other type of model.
{p 0 4}{cmdab:lap:lace} specifies that a bernouli or constant exposure poisson model should be estimated using the EM
laplace estimation. This option cannot be combined with the {cmd:od} option. This option is only available in HLM2.
{p 0 4}{cmd:l6} specifies that a bernouli model should be estimated using a 6-th order laplace approximation
to maximum likelihood. This option cannot be combined with the {cmd:od} option.
{p 0 4}{cmdab:ln:um:(}{it:integer}{cmd:)} specifies the maximum number of laplace iterations. The default is 50.
{p 0 4}{cmdab:maci:t:(}{it:integer}{cmd:)} specifies the maximum number of macro iterations in a non-linear model. In
a linear model, this option specifies the maximum number of iterations. The default is 100. If neither of the
{cmd:prompt} or {cmd:stop} options are specified, {cmd:macit:(}{it:integer}{cmd:)} is irrelevant, since the default is
to continue iteration until convergence.
{p 0 4}{cmdab:mici:t:(}{it:integer}{cmd:)} specifies the maximum number of micro iterations in a non-linear model.
The default is 14.
{p 0 4}{cmdab:macs:top:(}{it:#}{cmd:)} specifies the stopping criterion for macro iterations. The default is 0.0001.
{p 0 4}{cmdab:mics:top:(}{it:#}{cmd:)} specifies the stopping criterion for micro iterations in a non-linear model.
The default is 0.000001.
{p 0 4}{cmd:wgt1(}{it:varname}{cmd:)} specifies the level-1 design weight variable.
{p 0 4}{cmd:wgt2(}{it:varname}{cmd:)} specifies the level-2 design weight variable.
{p 0 4}{cmd:wgt2(}{it:varname}{cmd:)} specifies the level-3 design weight variable.
{p 0 4}{cmd:pwgt(}{it:varname}{cmd:)} specifies the level-1 precision weight variable..
{p 0 4}{cmd:fsig(}{it:#}{cmd:)} specifies that sigma-squared is to be set to {it:#}. Sigma-squared cannot be fixed in
an over-dispersed bernouli or poisson model.
{p 0 4}{cmdab:te:st} is used to specify multivariate hypothesis tests. The syntax for {it:test_specification} is
{p 4 8 2}{cmd:[}{it:test}{cmd:]}[{cmd:[}{it:test}{cmd:]}[{cmd:[}{it:test}{cmd:]}...]]
{p 4 4 2}Note: Distinguish between {cmd:[]}, which are to be typed, and [], which indicate optional arguments.
In addition, while not shown in the syntax diagram, brackets around {it:test} are only required with multiple test specifications.
{p 6 8 2}{it:test} is an expression of the form
{p 8 10 2}{it:eqn}[,{it:eqn}[,{it:eqn}...]]
{p 6 8 2}{it:eqn} is an expression of the form
{p 8 10 2}{it:exp}={it:exp}[={it:exp}...]
{p 6 8 2}{it:exp} is a linear combination of the variables and/or interactions in the model.
It can be any one of
{p 8 10 2}0{p_end}
{p 8 10 2}[{it:#}*][{it:var}][+|-[{it:#}*]{it:var}[+|-[{it:#}*]{it:var}...]]{p_end}
{p 6 8 2}{it:var} is any one of{p_end}
{p 8 10 2}{it:L1var}{p_end}
{p 8 10 2}{it:L2var}{p_end}
{p 8 10 2}{it:L3var}{p_end}
{p 8 10 2}{it:L2var}*{it:L1var}{p_end}
{p 8 10 2}{it:L3var}*{it:L1var}{p_end}
{p 8 10 2}{it:L3var}*{it:L2var}{p_end}
{p 8 10 2}{it:L3var}*{it:L2var}*{it:L1var}{p_end}
{p 4 4 2}For example, to test the null hypothesis that the coefficients on {it:var1} and {it:var2} are both equal to zero, one could write any of the following:{p_end}
{p 6 8 2}test([var1=0, var2=0]) {p_end}
{p 6 8 2}test([var1=var2=0]) {p_end}
{p 6 8 2}test([var1-var2=0, var2=0]) {p_end}
{p 4 4 2}However, to test the separate null hypotheses that the coefficient on {it:var1} equals zero and that the coefficient on {it:var2} equals zero, one could write the following: {p_end}
{p 6 8 2}test([var1=0][var2=0]) {p_end}
{p 0 4}{cmdab:dev:iance:(}{it:real integer}{cmd:)} specifies a deviance test.
The model will be compared to a model with deviance {it:real} and {it:integer}
degrees of freedom. Note that the deviance statistic and degrees of freedom
from previous models will be saved as e-class scalars (see {bf:stored results}
below), and so can be used in the {cmd:deviance} option to compare the
relative fit of different models.
{p 0 4}{cmdab:st:ore:(}{it:name}{cmd:)} specifies which set of estimates are to
be stored in e(b) and e(V). See also {cmd:robust} below. In {cmd:store}, {it:name}
may be {bf:lin} (used only with a linear model), {bf:us} (unit-specific estimates,
may be used with any non-linear model), {bf:pa} (population-average estimates,
may be used with any non-linear model), {bf:eml} (EM laplace estimates, may be used
when {cmd:laplace} option is specified}, or {bf:l6} (laplace-6 estimates, may be used
when {cmd:l6} option is specified}. If {cmd:store} is omitted, the default depends
on the model specified. If {cmd:laplace} is specified, the default is {bf:eml};
otherwise, if {cmd:l6} is specified, the default is {bf:l6}; otherwise, if the model
is any other non-linear model, the default is {bf:us}; finally, if the model is linear,
the default is {cmd:lin}.
{p 4 4}Note: The mechanism of storing the estimates is currently a bit clunky -- HLM
saves the estimates in files called "gamvcX.dat" and "tauvcX.dat", where X is
something like "us" or "eml" and so indicates which estimator the estimates
come from. In doing so, HLM writes over any existing files with these names,
but since not all files are produced by each run of HLM, there can be some
confusion regarding whether any existing "gamvcX.dat" come from the current
model. To deal with this, this program renames any existing "gamvcX.dat" and
"tauvcX.dat" files to some unused filenams like "gamvcX_tempN.dat", etc. before
HLM fits the current model. This program then reads the estimates from the newly
produced "gamvcX.dat" and "tauvcX.dat" files into matrices, stored in e(), and then
erases the current "gamvcX.dat" and "tauvcX.dat" files. Finally the
"gamvcX_tempN.dat" and "tauvcX_tempN.dat" files are renamed to their old
("gamvcX_tempN.dat" and "tauvcX_tempN.dat") names. In principal this works fine,
but if there is an error in the program and the hlm commands exits before completing,
it is possible that the "gamvcX_tempN.dat" and "tauvcX_tempN.dat" will not be
renamed to their original names. I tell you this so you won't wonder if a bunch of
"gamvcX_tempN.dat" and "tauvcX_tempN.dat" files start populating your hard drive.
{p 0 4}{cmd:robust} specifies that the variance-covariance matrix used to compute
the robust standard errors should be saved in e(V), when available (robust standard
errors are not available, for example, with laplace estimation).
{p 0 4}{cmd:noeq} suppresses the writing of the full HLM model equations in stata. The equations are written for
users to verify that the model is correctly specified; they do not affect the writing of the command file.
{p 0 4}{cmdab:lo:ng:} specifies that HLM should produce a full output file; the default is to produce only a reduced output file.
{p 0 4}{cmd:res1} specifies that HLM should produce a level-1 residual file.
The residual file is written to a file named {it:filename}_res1.dta, where
{it:filename} is the filename specified in the {cmdab:cmd:file:(}{it:newfile}{cmd:)}
command.
{p 0 4}{cmd:res2} specifies that HLM should produce a level-2 residual file.
The residual file is written to a file named {it:filename}_res2.dta, where
{it:filename} is the filename specified in the {cmdab:cmd:file:(}{it:newfile}{cmd:)}
command.
{p 0 4}{cmd:res3} specifies that HLM should produce a level-3 residual file.
The residual file is written to a file named {it:filename}_res3.dta, where
{it:filename} is the filename specified in the {cmdab:cmd:file:(}{it:newfile}{cmd:)}
command.
{p 0 4}{cmd:rv1(}{it:varlist}{cmd:)} specifies that the level-1 residual file
should include (in addition to the variables in the model, which are included by
default) the variables listed in {it:varlist}.
{p 0 4}{cmd:rv2(}{it:varlist}{cmd:)} specifies that the level-2 residual file
should include (in addition to the variables in the model, which are included by
default) the variables listed in {it:varlist}.
{p 0 4}{cmd:rv3(}{it:varlist}{cmd:)} specifies that the level-3 residual file
should include (in addition to the variables in the model, which are included by
default) the variables listed in {it:varlist}.
{p 0 4}{cmdab:ti:tle:(}{it:string}{cmd:)} specifies the title to be given to the HLM output file.
{p 0 4}{cmdab:gr:aph:(}{it:newfile}{cmd:)} specifies that HLM should write a graph equations (.geq) file
named {it:newfile}.
{marker par}
{bf:{dlgtab:options for hlm run}}
{p 0 4}{cmd:hlm2} is the default. It specifies that the .hlm command file is an
HLM2 command file.
{p 0 4}{cmd:hlm3} specifies that the .hlm command file is an HLM3 command file.
{p 0 4}{cmd:mdm(}{it:mdmfile}{cmd:)} specifies that {it:mdmfile} is the MDM file
to be used. If not specified, the MDM file specified in the most recent
{cmd:mdmdset} command is used.
{p 0 4}{cmdab:st:ore:(}{it:name}{cmd:)} specifies which set of estimates are to
be stored in e(b) and e(V). See also {cmd:robust} below. In {cmd:store}, {it:name}
may be {bf:lin} (used only with a linear model), {bf:us} (unit-specific estimates,
may be used with any non-linear model), {bf:pa} (population-average estimates,
may be used with any non-linear model), {bf:eml} (EM laplace estimates, may be used
when {cmd:laplace} option is specified}, or {bf:l6} (laplace-6 estimates, may be used
when {cmd:l6} option is specified}. If {cmd:store} is omitted, the default depends
on the model specified. If {cmd:laplace} is specified, the default is {bf:eml};
otherwise, if {cmd:l6} is specified, the default is {bf:l6}; otherwise, if the model
is any other non-linear model, the default is {bf:us}; finally, if the model is linear,
the default is {cmd:lin}.
{p 0 4}{cmd:robust} specifies that the variance-covariance matrix used to compute
the robust standard errors should be saved in e(V), when available (robust standard
errors are not available, for example, with laplace estimation).
{p 0 4}{cmd:noview} suppresses the viewing of the output file in the viewer window.
{marker par}
{title:Stored Results}
{p 0 0 2}Depending on the model specified, HLM will produce several sets of
estimated coefficients and the corresponding variance-covariance matrices. Each of
these are stored in Stata as e-class results. In addition, a number of scalars and
macros are srored:
{p 0 0 2}Scalars:
{col 6}e(sig2) {col 20}estimated level-1 residual variance (linear model only)
{col 6}e(N_1) {col 20}number of level-1 observations
{col 6}e(N_2) {col 20}number of level-2 observations
{col 6}e(N_3) {col 20}number of level-3 observations
{col 6}e(dev) {col 20}deviance statistic from linear model
{col 6}e(dev_l6) {col 20}deviance statistic from laplace-6 estimates
{col 6}e(dev_eml) {col 20}deviance statistic from EM laplace estimates
{col 6}e(df) {col 20}model df (number of parameters estimated)
{p 0 0 2}Macros:
{col 6}e(esttype) {col 20}type of estimates stored in e(b) and e(V)
{col 6}e(setype} {col 20}type of standard errors stored in e(V)
{col 6}e(cmdfile) {col 20}name of .hlm command file used
{col 6}e(depvar) {col 20}dependent variable
{col 6}e(model) {col 20}model type (distribution of outcome)
{col 6}e(cmd) {col 20}hlm2 or hlm3
{p 0 0 2}Matrices:
{col 6}e(b) {col 20}1 x K vector of estimated coefficients
{col 6}e(V) {col 20}K x K variance-covariance matrix
{col 6}e(b_lin) {col 20}1 x K vector of estimated coefficients from linear model
{col 6}e(b_us) {col 20}1 x K vector of estimated unit-specific coefficients
{col 6}e(b_pa) {col 20}1 x K vector of estimated population-average coefficients
{col 6}e(b_eml) {col 20}1 x K vector of estimated EM laplace coefficients
{col 6}e(b_l6) {col 20}1 x K vector of estimated laplace-6 coefficients
{col 6}e(V_lin) {col 20}K x K variance-covariance matrix, linear model
{col 6}e(V_us) {col 20}K x K variance-covariance matrix, unit-specific estimates
{col 6}e(V_pa) {col 20}K x K variance-covariance matrix, population-average estimates
{col 6}e(V_eml) {col 20}K x K variance-covariance matrix, EM laplace estimates
{col 6}e(V_l6) {col 20}K x K variance-covariance matrix, laplace-6 estimates
{col 6}e(V_r) {col 20}K x K robust variance-covariance matrix, linear model
{col 6}e(V_usr) {col 20}K x K robust variance-covariance matrix, unit-spec. estimates
{col 6}e(V_par) {col 20}K x K robust variance-covariance matrix, pop-average estimates
{col 6}e(se_lin) {col 20}1 x K standard error vector, linear model
{col 6}e(se_us) {col 20}1 x K standard error vector, unit-specific estimates
{col 6}e(se_pa) {col 20}1 x K standard error vector, population-average estimates
{col 6}e(se_eml) {col 20}1 x K standard error vector, EM laplace estimates
{col 6}e(se_l6) {col 20}1 x K standard error vector, laplace-6 estimates
{col 6}e(se_r) {col 20}1 x K robust standard error vector, linear model
{col 6}e(se_usr) {col 20}1 x K robust standard error vector, unit-spec. estimates
{col 6}e(se_par) {col 20}1 x K robust standard error vector, pop-average estimates
{col 6}e(tau2) {col 20}R x R level-2 tau matrix
{col 6}e(tau2_eml){col 20}R x R level-2 tau matrix, EM laplace estimates
{col 6}e(tau2_l6) {col 20}R x R level-2 tau matrix, laplace-6 estimates
{col 6}e(tau3) {col 20}R2 x R2 level-3 tau matrix
{col 6}e(invtaui) {col 20}inverse tau information matrix
{col 6}e(invtaui_l6) {col 20}inverse tau information matrix, laplace-6
{col 6}e(se_tau) {col 20}vector of standard errors of tau
{col 6}e(se_tau_l6) {col 20}vector of standard errors of tau, laplace-6
{p 0 0 2}The type of estimates stored in e(b) and e(V) are determined by the
{cmd:store} and {cmd:robust} options. In general, all available estimated
coefficients, variance-covariance matrices, and standard errors are stored, but
those stored in e(b) and e(V) will be available for postestimation commands such as
{help estout}, which can then be used to display results of multiple models.
{title:Examples}
{p 0 0 2}Suppose we have a data file with outcome variable {it:score}, level-one variables {it:age}, {it:sex}, {it:race1}, {it:race2}, {it:race3},
and level-two variable {it:private}. We first sort the data by the level-two id variable ({it:schoolid} here), and then make an
mdm file:
{p 4 8 2}sort schoolid {p_end}
{p 4 8 2}hlm mkmdm using new_mdm_file, id2(schoolid) l1(score age sex race1 race2 race3) l2(private) miss(now) {p_end}
{p 0 0 2}Next we tell Stata to use this mdm file in subsequent hlm models: {p_end}
{p 4 8 2}hlm mdmset new_mdm_file.mdm {p_end}
{p 0 0 2}Now, to estimate the following model: {p_end}
{p 4 0 0}score(ij) = B(0j) + B(1j)*age(ij) + B(2j)*sex(ij) + R(ij) {p_end}
{p 8 0 0} B(0j) = G(00) + G(01)*private(j) + U(0j) {p_end}
{p 8 0 0} B(1j) = G(10) {p_end}
{p 8 0 0} B(2j) = G(20) {p_end}
{p 0 0 2}we write the command:
{p 4 8 2}hlm hlm2 score int(int private rand) age(int) sex(int) rand, cmd(model01) run replace
{p 0 0 2}alternately, we could write the same command as:
{p 4 8 2}hlm hlm2 score int(int private rand) age sex rand, cmd(model01) run replace
{p 0 0 2}Now, to estimate the following model:
{p 4 0}score(ij) = B(0j) + B(1j)*race1(ij) + B(2j)*race2(ij) + B(3j)*race3(ij) + R(ij) {p_end}
{p 8 0} B(0j) = G(00) + G(01)*private(j) + U(0j) {p_end}
{p 8 0} B(1j) = G(10) + G(11)*private(j) {p_end}
{p 8 0} B(2j) = G(20) + G(21)*private(j) {p_end}
{p 8 0} B(3j) = G(30) + G(31)*private(j) {p_end}
{p 0 0 2}we write the command: {p_end}
{p 4 8 2}hlm hlm2 score int(int private rand) race1-race3(int private) rand, cmd(model02) run replace {p_end}
{p 0 0 2}Finally, to test the null hypothesis that the cofficients G(11), G(21), and G(31) are all equal to 0,
we would write: {p_end}
{p 4 8 2}hlm hlm2 score int(int private rand) race1-race3(int private) rand, cmd(model02) run replace test(private*race1=private*race2=private*race3=0) {p_end}
{title:Author}
{p 5 5}sean f. reardon{break}
Stanford University{break}
sean.reardon@stanford.edu {p_end}