How WesVar Works – Westat, Inc. (2023)

How does WesVar work?

WesVar Analysis Steps

Step 1 - Create a WesVar data file.

WesVar reads a variety of file formats. In Figure 1, the user reads data from an SPSS system file and provides the total sample weight and case ID variables. The data is a sample of cases from a national health and nutrition survey.

Figure 1. Point to the Data File screen to import files

In Figure 2, the user selects the remaining variables in the input file for further analysis. The file must then be saved as a WesVar data file.

Step 2 - Create new replication weights or import existing ones.

WesVar uses one of five replication methods to compute variances from search estimates. To do this, the user first creates replicated weights or imports existing replicated weights.

In Figure 3, the user provides one variable that defines the strata and another that defines the primary sampling units (UPAs). The user also selects a variance estimation method.

In this example, the sample design had two PSUs in each stratum. Balanced Repeat Replication (BRR) was specified and WesVar automatically chose an appropriately sized Hadamard array based on the number of layers in the data file.

Figure 3. Specifying variables to identify Stratum (VarStrat) and PSU (VarUnit). The BRR method of variance estimation is selected

In Figure 4, the replicated weights have been created and are labeled RPL01-RPL24. The user can also enter a different prefix in the Create Weights screen.

Figure 4. Using the BRR method to create replicated weights. On this screen the user saves the WesVar data file. In this case the name Demo.var was used

Step 3 - Create a WesVar workbook and specify your analysis.

The user defines analysis requirements in workbooks by selecting analysis types and variables with the mouse. In Figure 5, the user requested estimation of average diastolic and systolic blood pressure.

Figure 5. Defining calculated statistics for a table

Step 4 - Run the analysis and see the results.

The user then specifies that averages should be calculated for each cell of a RACE x GENDER table. The table also includes the estimated total number of cases (sum of weights/values) and row, column, and percentage totals for the estimated totals.

Figure 6. Specifying the variables that define a table

Figure 7 shows the output screen with mean diastolic and systolic blood pressure for three RACE categories (white and other, black and Mexican-American) crossed with GENDER. Row and column margins and overall averages are displayed.

Figure 7. Shows an output table of mean diastolic and systolic blood pressure by race and gender

Table ViewerUtility

Table Viewer is an easy-to-use tool for viewing, printing, and copying the output of one-way, two-way, and three-way tables previously created with WesVar 5. The Table Viewer presents WesVar's output in a grid which is easier to work with than the normal WesVar output, which produces a row for each cell of the user's table request.

To display a table in Table Viewer, the only files needed are name.wvb (the WesVar workbook file) and name.001, name.002, etc. (the table output files run in WesVar). WesVar data files and the WesVar software itself are not required to view the tables. Note that table lookups must be performed in WesVar to create the output files. Workbooks must be saved before the output can be viewed in TableViewer.

When installing WesVar, the Table Viewer program and some example files are placed in the directory C:\Programs\WesVar\tabview. To run Table Viewer, double-click the executable in that directory or go to Start - Programs - WesVar - Start TableViewer.exe.

RelabelUtility

The Relabel utility consists of a SAS program and a standalone executable program. Together they allow you to use information from a SAS format library to relabel values from a WesVar dataset that the user previously imported from SAS, since WesVar does not preserve labels when importing data.

When WesVar is installed, the Relabel SAS program, the standalone executable, and the Relabel instructions are placed in the C:\Programs\WesVar\relabel directory. To view the instructions, double-click ReadMe.html in that directory or go to Start - Programs - WesVar - RelabelInstructions.

WesVar-FAQ

A: Yes. Factor files have one record per replicate. The records must be in the same order as the physical order of the replicated weights in the input file used to create the data file (VAR). Suppose your replicated weights are named RWT1-RWT100 but sorted in alphanumeric order: RWT1, RWT10, RWT100, RWT11, etc. The factor files must also be in that order. Replicated weights are listed in their physical order in the data file on the Append Factors screen. Both the FPC and JKn factors must appear in the same order in the external files.
A: This message is issued in logistic regression when the convergence criterion for parameter estimates is not met within the specified maximum number of iterations. This can be due to the following reasons: 1. The convergence criterion has changed from a difference in log-likelihood in version 3 to a relative difference in log-likelihood in the current version. If you are a version 3 user, you may need to manually change the convergence criterion from 0.025 to 1E-6 in settings and in all version 3 workbooks imported into the current version. 2. There are few (if any) cases that have the property of being modeled in one cell. If your model includes a categorical independent variable, such as RACE, and there are no cases with the characteristic in the replicated subsample 43 for RACE.1 (dependent variable = 1), the estimation algorithm attempts to shift the parameter estimate to negative infinity. Therefore, the predicted probability for RACE.1 is 0. In this case, the parameter estimate does not converge, but becomes increasingly negative with each iteration. 3. You may need to reduce the categories of one of your independent variables to fix the problem. In the Regression Request/Output Control, check the Iteration History and Replication Coefficients check boxes for a detailed report of how the parameter estimates are changing.
A: An alternative is to use the following code to create your SAS transport files. Here is the code to create a SAS transport file (single member) named c:\export\wesvar.xpt from SD2 file c:\input\data.sd2: Options validvarname=v6; libname in1 'c:\input'; libname out1 xport 'c:\export\wesvar.xpt'; proc copy input=in1 output=out1; select dates; run; The directive "options validvarname=v6;" causes SAS to generate unique 8-character names for variables whose names are longer than 8 characters. The VALIDVARNAME option is valid only with SAS version 7 or later. It is not recognized (or required) by earlier versions of SAS.
A: SAS value labels are not imported into WesVar even if they were originally created in WesVar and exported to SAS. Use the relabel utility.
A: VARUNITS must be numbered sequentially starting at 1 in each VARSTRAT. The number of VARUNITS can differ from VARSTRAT to VARSTRAT.
A: WesVar only allows using the JKn method if there are at least two PSUs in each tier. To use the JKn method you would need to assemble each tier with a power supply with a different tier.
A: WesVar cannot read 64-bit SAS files. WesVar can only read 32-bit SAS files. The best way to ensure that the dataset is 32-bit is to specify the OUTREP=WINDOWS_32 dataset option when creating the dataset.
A: The US Government Configuration Baseline (USGCB) is a security baseline required by the Office of Management and Budget (OMB) to standardize the configuration of desktop computers used by US government agencies. WesVar has been tested for USGCB compliance on Windows 7® using a SCAP validated tool with USGCB scanner capability. All WesVar components are fully functional and working correctly as intended on systems using the USGCB on Windows 7. Default installation, operation, maintenance, updating and/or patching of WesVar components will not change the approved USGCB configuration settings. WesVar uses the Windows Installer service to install in the default Program Files directory and has the ability to install and uninstall unattended. All WesVar components can be run in the context of a standard user without elevated system administration rights.