R: Summarize Association Model Fits

summary.assoc {logmult}

R Documentation

Summarize Association Model Fits

Description

summary method for objects of class assocmod, including rc, rcL, rcL.trans, hmskew, hmskewL and yrcskew models.

Usage

## S3 method for class 'assocmod'
summary(object, weighting, ...)

## S3 method for class 'summary.assocmod'
print(x, digits = max(3, getOption("digits") - 4), ...)

Arguments

`object`	an association model of class `assocmod`.
`x`	an object of class `summary.gnm`.
`weighting`	what weights should be used when normalizing the scores.
`digits`	the number of siginificant digits to use when printing.
`...`	further arguments passed to `printCoefmat` by `print.summary.assocmod`, and currently ignored by `summary.assocmod`.

Details

print.summary.assocmod prints the original call to assoc; a summary of the deviance residuals from the model fit; the coefficients of interest of the model; the residual deviance; the residual degrees of freedom; the Schwartz's Bayesian Information Criterion value; the Akaike's An Information Criterion value.

Association coefficients are printed with their standard errors, p-values and significance stars. The “Normalized” columns contains normalized scores, i.e. their (weighted) sum is 0, their (weighted) sum of squares is 1, and their (weighted) cross-dimensional correlation is null. For models with only one layer (rc, hmskew, yrcskew), adjusted scores are printed in the “Adjusted” column: these correspond to normalized scores times the square root of the corresponding intrinsic association parameter (phi).

p-values correspond to normalized scores, and are computed under the assumption that estimators of coefficients are normally distributed, even if jackknife of bootstrap are used. See se.assoc for details about checking this assumption and the validity of jackknife or bootstrap results.

Note that setting the weighting argument to a value different from that used at the time of the fit discards the computed standard errors, if any.

Value

An object of class summary.assoc, with the following components:

`call`	the `call` component from object.
`diagonal`	the `diagonal` component from the object's `assoc` component.
`deviance.resid`	the deviance residuals, see `residuals.glm.`
`coefficients`	a matrix holding the association coefficients estimates, standard errors and p-values.
`diagonal`	a matrix holding the diagonal coefficients, if any.
`weighting`	the weigthing method used when normalizing the scores.
`deviance`	the `deviance` component from object.
`chisq`	the Pearson Chi-squared statistic for the model fit.
`dissim`	the dissimilarity index for the model fit.
`df.residual`	the `df.residual` component from object.
`bic`	the value of the BIC for the model fit (contrary to the value reported by `AIC` and `extractAIC`, the reference is 0 for the saturated model).
`aic`	the value of the AIC for the model fit (contrary to the value reported by `AIC` and `extractAIC`, the reference is 0 for the saturated model).
`family`	the `family` component from object.
`dispersion`	the estimated dispersion
`df`	a 3-vector of the rank of the model; the number of residual degrees of freedom; and number of unconstrained coefficients.

Author(s)

Milan Bouchet-Valat