Overview (MCONVERT command)

MCONVERT converts covariance matrix materials to correlation matrix materials, or vice versa. For MCONVERT to convert a correlation matrix, the matrix data must contain CORR values (Pearson correlation coefficients) and a vector of standard deviations (STDDEV). For MCONVERT to convert a covariance matrix, only COV values are required in the data.

Options

Matrix Files. MCONVERT can read matrix materials from an external matrix data file, and it can write converted matrix materials to an external file.

Matrix Materials. MCONVERT can write the converted matrix only or both the converted matrix and the original matrix to the resulting matrix data file.

Basic Specification

The minimum specification is the command itself. By default, MCONVERT reads the original matrix from the active dataset and then replaces it with the converted matrix.

Syntax Rules

  • The keywords IN and OUT cannot specify the same external file.
  • The APPEND and REPLACE subcommands cannot be specified on the same MCONVERT command.

Operations

  • If the data are covariance matrix materials, MCONVERT converts them to a correlation matrix plus a vector of standard deviations.
  • If the data are a correlation matrix and vector of standard deviations, MCONVERT converts them to a covariance matrix.
  • If there are multiple CORR or COV matrices (for example, one for each grouping (factor) or one for each split variable), each will be converted to a separate matrix, preserving the values of any factor or split variables.
  • All cases with ROWTYPE_ values other than CORR or COV, such as MEAN, N, and STDDEV, are always copied into the new matrix data file.
  • MCONVERT cannot read raw matrix values. If your data are raw values, use the MATRIX DATA command.
  • Split variables (if any) must occur first in the file that MCONVERT reads, followed by the variable ROWTYPE_, the grouping variables (if any), and the variable VARNAME_. All variables following VARNAME_ are the variables for which a matrix will be read and created.

Limitations

  • The total number of split variables plus grouping variables cannot exceed eight.