Prints the results of a generalized linear model fit from the glim
function in a table format.
USAGE:
glim.print(glim.list, digits=4, scale=F)
REQUIRED ARGUMENTS:
glim.list:
a list containing the result of a fit from the glim function.
OPTIONAL ARGUMENTS:
digits:
number of significant digits to print for the coefficients and standard
errors.
scale:
a logical value to determine if the deviance should be scaled using the value
of the scale component in glim.list.
If TRUE, then increments in the deviance are tested using an F test.
If FALSE, the test is done using the Chi-square distribution.
If weights were used in the original fit, neither distribution may be correct.
SIDE EFFECTS:
a table is printed containing the coefficients, their standard errors,
their studentized values, and p-values for their significance.
The significance of adding the variables in the sequence of models is also
tested.
The coefficients given in the table are for the final model, not for
the incremental models.
DETAILS:
The p-values for the coefficients are from a two-sided Z test (the studentized
coefficients are assumed to be Normally distributed).
When scale=FALSE, Chi-square tests are performed on the difference
in deviance between adjacent models in the sequence.
The difference in deviance is divided by the final scale estimate and an
F test is performed when scale=TRUE.
The null hypothesis for the tests is that the added covariates do not
help predict the response.
These tests only use an approximation to the distribution of the deviance
so the p-values should be used as a guide to modeling rather than taken
literally.
The Chi-square and F tests are, however, more reliable than the
Z test for coefficients.