multilingual: Multilingual Free Responses Data

In pamctdp: Principal Axes Methods for Contingency Tables with Partition Structures on Rows and Columns

Description

The data are extracted from a large international survey (Hayashi et al., 1992). People from four countries (Great Britain, France, Italy, Japan) are asked several closed questions and, moreover, the open-ended question: “What is the most important thing to you in life?” is considered. The Japanese an- swers are romanized. In each country, the free answers are grouped into 18 category-documents by crossing gender (male, female), age (into three categories: 18-34, 35-54, 55 and over) and educational level (into three categories: low, medium and high). Then, for each country, from the count of words in the whole answers, the lexical table arises by crossing the 18 documents and the most frequent words. Only the words used at least 20 times are kept.

Usage

data(multilingual)

Format

multilingual is a list with components.

tab

is a data frame containing frequency table: 18 rows (categorical variable = gender x age x educational level)

rbl

is a vector containing the repartition of words for the 4 countries

tab3

concatenation of three frequency tables (gender, age, educational level)

cbl3

is a vector containing the category numbers in each table

Source

Hayashi et al.,1992. We thank Profesor Lebart to put at our disposal this data set.

References

M. Becue and J. Pages and C.E. Pardo (2004). Analysis of cross-language open-ended questions through MFACT, in: Classification, Clustering, and Data Mining Applications. EditorsDavid Banks,Leanna House, Frederick R. McMorris, Phipps Arabie and Wlfgang Gaul. Heidelberg. pringer". pages 553-561. The 2004 Meeting of the International Federation of Classification Societies. Illinois Institute of Technology in Chicago, 15-18 July 2004 See in directory: BecuePagesPardo04.pdf.

Examples

# MFACT with pamctdp functions
data(multilingual)
# simple correspondence analysis
sca <- dudi.coa(multilingual$tab,scannf=FALSE,nf=2)
# MFACT analysis 
mfact <- witwit.model(sca,multilingual$rbl,18,weight="mfa",scannf=FALSE,nf=2)
inertia(mfact)$row
# MFACT = ACIBP*homotecia
wibca2mfa(mfact)
# plot of texts
plot(mfact,Trow=FALSE,cframe=0.5)
# plot of words with representation quality on the first plan >= 40%
dev.new()
plot(mfact,ucal=40,all.point=FALSE)
# partial coordinates
parmfact <- partial.wwm(mfact)
#superimposed representation of categorias with age between 35 and 54
#1. points selection
age2 <- names(multilingual$tab)[substr( names(multilingual$tab),3,4)=="35"]
#2. new graphics window
dev.new(width=6,height=8)
#3. superimposed representation
# clic in global points and clic in the top to finish
# plot(parmfact,graph="cols",coleti=age2)

pamctdp documentation built on May 1, 2019, 10:19 p.m.