.TH REGRESS 1STAT "October 16, 1980"
.SH NAME
regress \- statistics: multivariate linear regression and correlation
.SH SYNOPSIS
.B regress
[column names]
.SH DESCRIPTION
.I corr/regress
performs a general linear correlation analysis
with an optional multiple linear regression analysis.
.I corr/regress
reads from the standard input (via redirection with < or piped
with |) and writes to the standard output.
The program is based on the methods described by
Kerlinger and Pedhazur (1973) in
.I "Multiple Regression in Behavioral Research."
.PP
.I corr/regress
prints means, standard deviations,
and correlations for up to ten variables.
It has two names, corr and regress, and depending on which you
use, you will get different amounts of information.
If the program is called "corr", then only the correlation analysis is done.
If the program is called "regress", then
regression equation coefficients (slopes and intercepts), the squared multiple
correlation coefficients and significance tests are computed
for every possible dependent variable.
.I "The regression coefficients are printed in the column corresponding to each variable.
.PP
The program assumes that its input is a file of lines,
each containing an equal number of numerical fields.
Optionally, names for these fields can be supplied in your call to
the program, but if none are specified, A, B, C, etc. are used.
.SH DIAGNOSTICS
.I regress
will complain about "ragged input"
if there are not an equal number of fields on each line.
If some fields are not numerical,
.I regress
will complain about "non-numerical input."
If
.I regress
tries to do a regression analysis,
and finds that two variables are perfectly correlated,
.I regress
will complain about a "non-singular correlation matrix."
.SH SEE\ ALSO
dm(1CSL), biplot(1CSL), pair(1CSL), abut(1CSL)
.br
G. Perlman,
Data analysis programs for the UNIX operating system,
.I "Behavior Research Methods and Instrumentation,"
1980,
.I 12,
554-558.
.SH AUTHOR
Gary Perlman
