IBM Support

Can I compute "robust standard errors" in SPSS?

Troubleshooting


Problem

I want to use White's heteroskedascity-consistent standard errors in a linear regression model. Can SPSS compute these robust standard errors?

Resolving The Problem

The GENLIN procedure for generalized linear models and generalized estimating equations offers robust standard errors. These are the original Huber-White estimators for linear models. To access the GENLIN procedure in the menus, click on Analyze>Generalized Linear Models>Generalized Linear Models for a standard model with independent observations, or Analyze>Generalized Linear Models>Generalized Estimating Equations for a GEE model for correlated observations. Specification of the robust covariance matrix estimator is done on the Estimation tab for a generalized linear model. For a GEE model, the robust covariance matrix estimator is the default, and is specified on the Repeated tab. The newer GENLINMIXED procedure (Analyze>Mixed Models>Generalized Linear) offers similar capabilities.

The CSGLM, CSLOGISTIC and CSCOXREG procedures in the Complex Samples module also offer robust standard errors. The methods used in these procedures provide results similar to Huber-White or sandwich estimators of variances with a small bias correction equal to a multiplier of N/(N-1) for variances. To obtain these variance estimates, compute a constant sampling weight variable with a value of 1 for each case, and define your design as a single stage design without strata or clusters, using WR (with replacement) estimation:

1) Compute a constant variable with a value of 1 for each case.

2) In the menus, click Analyze>Complex Samples>Prepare for Analysis.

3) In the first panel of the wizard, you want the default option, which is to create a plan file. Specify a name for your file and click Next.

4) At the following panel, move the constant variable you've just created into the Sample Weight slot. Click Finish.

5) From the menus, select Analyze>Complex Samples>General Linear Model.

6) The plan file you just created should be selected at the top of the dialog box. If you are revisiting later, you may need to reselect this file. Then click Continue.

7) Specify the dependent variable and your predictors. Click Statistics and select the desired statistics, deselecting any defaults you do not want to see. Click Continue.

8) If you want to get any custom hypothesis tests or other output, click on appropriate buttons and specify these; otherwise, or afterwards, click OK.

[{"Product":{"code":"SSLVMB","label":"IBM SPSS Statistics"},"Business Unit":{"code":"BU059","label":"IBM Software w\/o TPS"},"Component":"Not Applicable","Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"Not Applicable","Edition":"","Line of Business":{"code":"LOB10","label":"Data and AI"}}]

Historical Number

38011

Document Information

Modified date:
16 April 2020

UID

swg21477323