sea.table {ldsa}R Documentation

Compute an Entailment Table

Description

Computes a descriptive entailment table for an input data set, much in the style of White (2000): Douglas R. White, Manual for Statistical Entailment Analysis. World Cultures 11(1):77-90.

Abstract of White (2000). A programmed statistical method developed for the analysis of binary data by the author explicates how to find approximations to discrete Boolean relations of inclusion, mutual exclusion, and collective exhaustion that satisfy empirical conditions for transitivity, and thus which facilitate formulation of rules and generalizations in discrete form ("If ... then ...") that are also logically transitive. Signal detection methods are used to reject relationships that could be due to chance by comparing actual relationships to those found in Monte Carlo simulations of comparable random datasets. The analytic results constitute a discrete network structure of nontrivial empirical implications that characterize a dataset.

Usage

sea.table(dat)

Arguments

dat a data.frame with observations on rows; missing data is permitted.

Details

Examination of an entailment table is a useful first step in conducting an entailment analysis; additional measures such as local tests or confirmatory latent class analysis can then be applied to discriminate among possible models. See sea.entailment for a description of entailment types, and their respective falsification conditions.

Value

An object (data frame) of class sea.table, containing the following columns (note that each row corresponds to a variable pair):

xnam Name of the X variable for this pair
ynam Name of the Y variable for this pair
Corr X,Y correlation for this pair
R^2 Squared X,Y correlation for this pair
XYErr Error rate for X=>Y
XYErrZ Z-score for error cell associated with X=>Y
XYErrp p-value for a one-sided exact binomial test of the error rate associated with X=>Y, versus the marginal probability
XYStr Does X=>Y satisfy White's “strong relationship” criterion?
YXErr Error rate for Y=>X
YXErrZ Z-score for error cell associated with Y=>X
YXErrp p-value for a one-sided exact binomial test of the error rate associated with Y=>X, versus the marginal probability
YXStr Does Y=>X satisfy White's “strong relationship” criterion?
XnYErr Error rate for X=>!Y
XnYErrZ Z-score for error cell associated with X=>!Y
XnYErrp p-value for a one-sided exact binomial test of the error rate associated with X=>!Y, versus the marginal probability
nXYErr Error rate for !X=>Y
nXYErrZ Z-score for error cell associated with !X=>Y
nXYErrp p-value for a one-sided exact binomial test of the error rate associated with !X=>Y, versus the marginal probability
N00 Number of X=0,Y=0 observations
N01 Number of X=0,Y=1 observations
N10 Number of X=1,Y=0 observations
N11 Number of X=1,Y=1 observations

Author(s)

Carter T. Butts buttsc@uci.edu

References

White, D. R. 2000. Manual For Statistical Entailment Analysis 2.0: Sea.Exe World Cultures, 11(1): 77-90.

See Also

sea.enumerate, sea.entailment, sea.net, binom.test


[Package ldsa version 0.1-2 Index]