Output datasets created by parmest and parmby
The output dataset (or resultsset) created by parmest or parmby has one observation per parameter of a set of estimation results, or one observation per parameter per by-group if the by() option is specified with parmby. If the rename() option is not specified, then the output dataset contains the following variables:
Default name Description -------------------------------------------------------------------------- idnum Numeric dataset ID idstr String dataset ID eq Equation name parm Parameter name msetype Parameter matrix stripe element type omit Parameter omit status empty Parameter empty cell status label Parameter label ylabel Y-variable label estimate Parameter estimate stderr SE of parameter estimate dof Degrees of freedom t t-test statistic z z-test statistic p P-value stars Stars for P-value minyy Lower xx% confidence limit maxyy Upper xx% confidence limit em_y yth macro estimation result specified by emac() es_y yth scalar estimation result specified by escal() ev_y yth vector estimation result specified by evec() er_y_k kth row of yth estimation result specified by erows() ec_y_k kth column of yth estimation result specified by ecols() --------------------------------------------------------------------------
The variables idnum and idstr are only present if the user specifies the options idnum() and idstr(), respectively. The variable eq is only present if the estimation results are from a command which creates estimation matrices with equation names. The variables msetype, omit, empty, label and ylabel are only present if the user specifies the options msetype, omit, empty, label and ylabel, respectively. Either the variables dof and t are present or the variable z is present, depending on whether the confidence limits and P-values are calculated using the t-distribution or the standard Normal distribution. The variable dof is calculated as specified by the dof() and dfmatrix() options, or from the estimation result e(df_r), if dof() and dfmatrix() are not specified. Note that the degrees of freedom in dof may be different for different parameters of the same model, and are not always integers. The P-values in the variable p test the hypothesis that the corresponding parameter is zero, or one if the eform option is specified. The variable stars is only present if the option stars() is specified, and contains the specified numbers of stars for the P-values in the variable p. The xx% confidence limits minyy and maxyy are calculated using a list of one or more confidence levels xx, which may be specified in the level() option, or by set level, and which is a single confidence level of 95% if not specified. The number yy used to number the xx percent confidence limits minyy and maxyy is equal to the confidence level xx unless the user specifies clnumber(rank), in which case the number yy is the rank, in ascending order, of the confidence level xx. There is one em_y variable for each macro estimation result in the list specified by the emac() option, one es_y variable for each scalar estimation result specified by the escal() option, one ev_y variable for each vector estimation result specified by the evec() option, one er_y_k variable for each row of each matrix estimation result specified by the erows() option, and one ec_y_k variable for each column of each matrix estimation result specified by the ecols() option. The variables em_y are string variables, truncated if necessary to the maximum length for a string variable under the edition of Stata used. All of these variables can be renamed using the rename() option.
The output dataset (or resultsset) created by parmby contains all the variables in the output dataset created by parmest, and also the following additional variables:
Default name Description -------------------------------------------------------------------------- by-variables Variables specified in the by() option parmseq Parameter sequence number command Estimation command --------------------------------------------------------------------------
The by-variables are only present if the by() option is specified. The variable command is only present if the command option is specified. The parmby output dataset is sorted primarily by the by-variables, in the order specified in the by() option, if the by() option is specified, and sorted secondarily by parmseq, which contains the sequential order of the parameter in the estimation result vector (which is usually e(b}). The variables command and parmseq can be renamed by parmby using the rename() option, but the by-variables cannot be renamed by parmby.
The eform, dof(), level() and clnumber() options are described in parmest_ci_opts. The idnum(), idstr(), label, ylabel, stars(), emac(), escal(), evec(), erows() and ecols() options are described in parmest_varadd_opts. The rename() option is described in parmest_varmod_opts. The by() and command options are described in parmby_only_opts.
Saved characteristics
The variables minyy and maxyy, containing the xx% confidence limits, all have the variable characteristic level, containing the respective percentage confidence levels xx specified for these variables by the level() option.
Author
Roger Newson, Imperial College London, UK. Email: r.newson@imperial.ac.uk
Also see
Manual: [U] 20 Estimation and postestimation commands
Help: [U] 20 Estimation and postestimation commands parmest, parmby, parmest_outdest_opts, parmest_ci_opts, parmest_varadd_opts, parmest_varmod_opts, parmby_only_opts