Overview (GLM: Repeated Measures command)

This section discusses the subcommands that are used in repeated measures designs, in which the dependent variables represent measurements of the same variable (or variables) taken repeatedly. This section does not contain information on all of the subcommands that you will need to specify the design. For some subcommands or keywords not covered here, such as DESIGN, see GLM: Univariate. For information on optional output and the multivariate significance tests available, see GLM: Multivariate.

  • In a simple repeated measures analysis, all dependent variables represent different measurements of the same variable for different values (or levels) of a within-subjects factor. Between-subjects factors and covariates can also be included in the model, just as in analyses not involving repeated measures.
  • A within-subjects factor is simply a factor that distinguishes measurements made on the same subject or case, rather than distinguishing different subjects or cases.
  • GLM permits more complex analyses, in which the dependent variables represent levels of two or more within-subjects factors.
  • GLM also permits analyses in which the dependent variables represent measurements of several variables for the different levels of the within-subjects factors. These are known as doubly multivariate designs.
  • A repeated measures analysis includes a within-subjects design describing the model to be tested with the within-subjects factors, as well as the usual between-subjects design describing the effects to be tested with between-subjects factors. The default for the within-subjects factors design is a full factorial model which includes the main within-subjects factor effects and all their interaction effects.
  • If a custom hypothesis test is required (defined by the CONTRAST, LMATRIX, or KMATRIX subcommands), the default transformation matrix (M matrix) is taken to be the average transformation matrix, which can be displayed by using the keyword TEST(MMATRIX) on the PRINT subcommand. The default contrast result matrix (K matrix) is the zero matrix.
  • If the contrast coefficient matrix (L matrix) is not specified, but a custom hypothesis test is required by the MMATRIX or the KMATRIX subcommand, the contrast coefficient matrix (L matrix) is taken to be the L matrix which corresponds to the estimable function for the intercept in the between-subjects model. This matrix can be displayed by using the keyword TEST(LMATRIX) on the PRINT subcommand.

Basic Specification

  • The basic specification is a variable list followed by the WSFACTOR subcommand.
  • Whenever WSFACTOR is specified, GLM performs special repeated measures processing. The multivariate and univariate tests are provided. In addition, for any within-subjects effect involving more than one transformed variable, the Mauchly test of sphericity is displayed to test the assumption that the covariance matrix of the transformed variables is constant on the diagonal and zero off the diagonal. The Greenhouse-Geisser epsilon and the Huynh-Feldt epsilon are also displayed for use in correcting the significance tests in the event that the assumption of sphericity is violated.

Subcommand Order

  • The list of dependent variables, factors, and covariates must be first.

Syntax Rules

  • The WSFACTOR (within-subjects factors), WSDESIGN (within-subjects design), and MEASURE subcommands are used only in repeated measures analysis.
  • WSFACTOR is required for any repeated measures analysis.
  • If WSDESIGN is not specified, a full factorial within-subjects design consisting of all main effects and all interactions among within-subjects factors is used by default.
  • The MEASURE subcommand is used for doubly multivariate designs, in which the dependent variables represent repeated measurements of more than one variable.

Limitations

  • Any number of factors can be specified, but if the number of between-subjects factors plus the number of split variables exceeds 18, the Descriptive Statistics table is not printed even when you request it.
  • Maximum of 18 within-subjects factors.
  • Memory requirements depend primarily on the number of cells in the design. For the default full factorial model, this equals the product of the number of levels or categories in each factor.