GSS93: Political Orientation and Religion in the United States in...

In cmm: Categorical Marginal Models

Political Orientation and Religion in the United States in 1993 (General Social Survey, 1993)

Description

Self-reported Political Orientation (P), Religion (R), and Opinion of Teenage Birth-control (B) of a sample of 911 US citizens in 1993. The data stem from the General Social Survey. The data are tabulated in Bergsma, Croon, and Hagenaars (2009, Table 2.1, Table 2.3). See Section~2.1 in Bergsma, Croon, and Hagenaars (2009). Several worked examples involving this data set are listed below but more can be found at http://stats.lse.ac.uk/bergsma/cmm/R files/GSS93.R

Usage

data(GSS93)

Format

A data frame with 911 observations on the following three variables.

P

Political orientation (ordered): 1 = Extremely liberal; 2 = Liberal; 3 = Slightly liberal; 4 = Moderate; 5 = Slightly conservative; 6 = Conservative; 6 = Extremely conservative.

R

Religion (factor): 1 = Protestant; 2 = Catholic; 3 = Other.

B

Opinion of birth control (ordered): 1 = Strongly agree; 2 = Agree; 3 = Disagree; 4 = Strongly disagree;

Source

General Social Survey (1993)

References

Bergsma, W. P., Croon, M. A., & Hagenaars, J. A. P. (2009). Marginal models for dependent, clustered, and longitudinal categorical data. New York: Springer

General Social Survey (1993).

Examples

data(GSS93)

## Table 2.1 of Bergsma, Croon, & Hagenaars (2009)
addmargins(table(GSS93[,1:2]))

## Table 2.2 of Bergsma, Croon, & Hagenaars (2009)
# Specify coefficients
coeff <- list("log",diag(21))
SampleStatistics(dat = GSS93[, 1 : 2],
 coeff = coeff,
 CoefficientDimensions = c(7, 3),
 Labels = c("P","R"), 
 ShowParameters = TRUE, 
 ShowCoefficients = FALSE)

## Table 2.3 of Bergsma, Croon, & Hagenaars (2009)
ftable(B + R ~ P , data = GSS93)


########################################################
## Models for table PR

#independence of P and R
bt1 <- ConstraintMatrix(c("P", "R"), list(c("P"), c("R")), c(7,3));
#linear by nominal model
bt2 <- ConstraintMatrix(var = c("P", "R"), 
  suffconfigs = list(c("P", "R")), 
  dim = c(7, 3), 
  SubsetCoding = list(c("P", "R"), c("Linear", "Nominal")))
#linear by nominal model with equality of first two nominal parameters
bt3 <- ConstraintMatrix(var = c("P", "R"), 
 suffconfigs = list(c("P", "R")), 
 dim = c(7, 3), 
 SubsetCoding = list(c("P", "R"), list("Linear", rbind(c(1, 1, 0), c(0, 0, 1)))))
m <- MarginalModelFit(dat = GSS93[,1:2],
 model = list(bt2,"log"), 
 ShowCoefficients = FALSE, 
 ShowProgress = 1, 
 ShowParameters = TRUE, 
 CoefficientDimensions = c(7, 3),
 Labels = c("P", "R"),
 ParameterCoding = list("Polynomial", "Effect"))


########################################################
## Models for table PRB

#various loglinear models
bt1 <- ConstraintMatrix(c("P", "R","B"), list(c("P","R"),c("B")), c(7,3,4))
bt2 <- ConstraintMatrix(c("P", "R","B"), list(c("P","R"),c("R","B")), c(7,3,4))
bt3 <- ConstraintMatrix(c("P", "R","B"), list(c("P","R"),c("P","B")), c(7,3,4))
bt4 <- ConstraintMatrix(c("P", "R","B"), list(c("P","R"),c("P","B"),c("R","B")), c(7,3,4))
bt5 <- ConstraintMatrix(c("P", "R","B"), list(c("P","R"),c("P","B"),c("R","B")), c(7,3,4), 
 SubsetCoding = list(list(c("P", "B"), c("Linear", "Linear")), 
                list(c("R", "B"), c("Nominal", "Linear"))))

m <- MarginalModelFit(dat = GSS93,
 model = list(bt2,"log"), 
 ShowCoefficients = FALSE, 
 ShowProgress = 1, 
 ShowParameters = TRUE,
 CoefficientDimensions =c(7, 3, 4),
 Labels = c("P", "R", "B"), 
 ParameterCoding = list("Polynomial", "Polynomial", "Effect"))

cmm documentation built on Aug. 10, 2023, 1:07 a.m.