*
Note: These are older
DOS-only programs
These programs were wrtten by Sean Hammond at the University of Cork .. they might still be of value to some and so are included for posterity... the MSA within the facet package is particularly interesting !
CFA: Type
Identification by Configural Frequency Analysis Click
here
to download the zipped program. The zipped file is 44k in size and is named CFA.ZIP.
This DOS program will apply zero, first and second order configural frequency
analyses on data in the manner presented in Krauth and Leinert (1973). A good
English text on CFA and its more recent developments is von Eye (1990). The user
provides the data to the program in the form of a data file on disk. This file
may be in one of two forms (Raw Data or Profile Table).
Running the Program
The program is run by entering the instruction CFA at the DOS prompt. CFA is
menu driven and asks the user to supply information interactively. It asks the
following questions:-
1. The name of the data file.
2. The name of the file to contain the results.
3. The form of the data file (T for Table, R for Raw).
4. The number of variables to be analysed (3 in our example data sets).
5. The format of the profiles. This is the FORTRAN convention using I notation.
The format statement MUST be enclosed in parenthesis.
The output table is the standard CFA output as exemplified by Krauth and Lienert
(1973), Lautsch and von Weber (1990), von Eye (1990) and Krauth 1993). A
readme.doc file accompanies the programs.
References:
Krauth J. (1993) Einfuhrung in die
Konfigurationsfrequenzanalyse. Weinheim: BELTZ
Krauth J. and Lienert G.A. (1973) Die
Konfigurationsfrequenzanalyse und ihre Anwendung in Psychologie und Medizin.
Munich: Alber.
Lautsch E. and von Weber S (1990) Die
Konfigurationsfrequenzanalyse: Methoden und Anwendung. Berlin: Volk und Wissen.
von Eye A. (1990) Introduction to Configural Frequency
Analysis: The search for types and antitypes in cross-classifications.
Cambridge: University of Cambridge Press.
Author: Sean Hammond, University of
Cork, Ireland
CORRES: A Program
for the Correspondence Analysis of Contingency Tables Click
here
to download the zipped program. The zipped file is 100k in size and is named CORRES.ZIP.
CORRES is a programme designed to perform a correspondence analysis of data
tables. This procedure is part of a whole family of data analysis techniques
known disparately as Dual Scaling, Homals, MSA, Biplot, and h-Plot analysis.
They owe much to the pioneering work of psychometricians such as Guttman,
Eckhart, Young and Hotelling. The user of CORRES is urgently advised to peruse
at least one of the following texts:
Benzecri J.P. (1973) L'Analyse des Données. Vol 2. Dunod;
Paris
Gifi A. (1990) Non-Linear Multivariate Analysis. Dept of Data
Theory; University of Leiden
Greenacre M.J. (1984) Theory and Applications of Correspondence
Analysis. Academic Press; London.
Lebart L., Morineau A. and Warwick K.M. (1984) Multivariate
Descriptive Statistical Analysis. Wiley & Sons; New York.
Nishisato S. (1980) Analysis of Categorical Data: Dual Scaling
and its Applications. University of Toronto Press; Toronto.
Seber G.A.F. (1984) Multivariate Observations. Wiley &
Sons; New York.
CORRES is a DOS programme written for all IBM compatible computers and Epson
compatible printers. It may be run on all configurations from Floppy to hard
disk systems. If an 8087 version is required, this may be provided by the
author.The languages in which this programme was written are QBASIC and PASCAL
with one routine written in FORTRAN77. A readme.doc file accompanies the
programs.
Author: Sean Hammond, University of
Cork, Ireland
FACET: A suite of
programs to support Facet Analysis - Smallest Space Analysis, Multidimensional
Scalogram Analysis, Partial Order Scalogram Analysis Click
here
to download the zipped program. The zipped file is 294.5k in size and is named FACET.ZIP.
The programs run in the DOS environment.
MULTIDIMENSIONAL SCALING ANALYSIS (SSA)
The multidimensional scaling module in PAP utilises the non-metric algorithm
proposed by Guttman (1968) which he termed smallest space analysis. Unlike the
powerful MINISSA algorithm PAP (the Hammond Psychometric Analysis Package)
offers a single-phase algorithm using Guttman's (1968) permutation procedure.
This is one of the three modules in PAP which may be of particular value to
those wishing to utilise non-metric data analysis procedures, especially those
whose data derives from a Facet Theory framework. When PAP passes control to the
MDS module the user offered the option of outputing the variable statistics and
the inter-variable similarity matrix. The user is then asked what the minimum
dimensionality for the analysis is to be. Next the maximum dimensionality is
requested. The multidimensional scaling analysis then proceeds. First the
minimum dimensionality is fitted to the original data and then the next highest
dimensionality until the maximum dimensionality has been fitted. At the end of
each analysis the user is given the option of outputting the derived distance
matrix. The user is then offered the option of viewing the n-dimensional plot on
the VDU screen.
The output from the MDS module includes:-
1. Descriptive statistics for variables.
2. The inter-variable correlation matrix.
3. For each solution:-
Before using the SSA analysis the user is advised to consult one or more of
the following references:-
Bailey K.D. (1974) Interpreting Smallest Space Analysis.
Sociological Methods and Research, 3: 3-29.
Coxon A.P.M. (1982) The User's Guide to Multidimensional
Scaling. Heinemann Educational Books: London.
Guttman L. (1968) A general non-metric technique for finding
the smallest coordinate space for a configuration of points. Psychometrika, 33:
469-506.
Schlesinger I.M. and Guttman L. (1969) Smallest Space Analysis
of intelligence and achievement tests. Psychological Bulletin, 71: 95-100
MULTIPLE SCALOGRAM ANALYSIS (MSA)
The Multiple Scalogram Analysis (MSA) of PAP is based upon the algorithm
developed by Lingoes (1979). It is designed to operate upon data measured in
nominal categories. Zeros are assumed to represent missing data. This module
estimates the weights which produce a maximally reliable composite scale of
items measured in nominal categories. The user may specify the number of
dimensions to be identified. The method is a non- metric extension of Guttman's
Least Squares technique for scalogram analysis. The widest use of MSA is to
provide the analyst with an n-dimensional representation of the similarities
between profiles (cases). When PAP transfers control to MSA the user is asked to
provide the dimensionality to be examined. No further prompting is given.
The output from the MSA module includes:-
1. A table of unique profiles.
2. A frequency distribution table for the categories of each variable.
3. A table of weights for each profile on each dimension.
4. A table for each variable giving the category weights for each dimension.
5. A table of profile co-ordinates.
6. The outer-point matrix.
7. Plots of the profiles in n-dimensional space.
8. Optional superimposed plots for each item including outer-point boundary
demarcation.
Before using the MSA module users are advised to consult with one or more of
the following references:-
Guttman L. (1950) The basis for Scalogram Analysis. In
(Stouffer S.S. Measurement and Prediction). Princeton: Princeton University
Press.
Guttman L. (1985) Multidimensional Structuple Analysis (MSA-I)
for the Classification of Cetacea. In (Marcotorchino J.F., Proth J.M. and
Janssen J. Data Analysis in Real Life Environment). North Holland: Amsterdam.
Lingoes J.C. (1979) The Multivariate Analysis of
Qualitative Data. In (Lingoes J.C., Roskam E.E. and Borg I. Geometric
Representations of Relational Data). Mathesis Press: Ann Arbor
Zvulun E. (1978) Multidimensional Scalogram Analysis: The Method and its Application. In (Shye S. Theory Construction and Data
Analysisin the Behavioral Sciences). Jossey-Bass: San-Francisco
PARTIAL ORDER SCALOGRAM ANALYSIS (POSA)
The Partial Order Scalogram Analysis provided by PAP is derived from the
algorithm developed by Shye (1978, 1986). Although the technique is quite
straight forward the computational effort in performing the analysis is immense.
Therefore the programme may take a while to complete the analysis. The time
taken is a direct function of the number of items, the number of people and the
operating speed of the computer you are using. Throughout the analysis, POSA,
like all other modules in PAP, will inform the user of its status at the VDU.
Output from POSA includes:-
1. An ordered table of person profiles. Extreme profiles are added by the
programme if absent in the data.
2. An inter-item matrix of monotonicity coefficients (Guttman's Mu or Yules Q).
3. Model fitting indices:-
4. A table of partial order co-ordinates consisting of base and transformed
co-ordinates.
5. A table of weak monotonicity coefficients between items and the scale models.
6. A 2-dimensional plot of the profiles.
7. An optional superimposed plot of each item.
Before using the POSA the user is advised to consult one or more of the
following references:-
Guttman L. and Levy S. (1985) The Partial Order of Traffic
Conflicts by Severity. In (Marcotorchino J.F., Proth J.M. and Janssen J.
Data Analysis in Real Life Environment: Ins and Outs of Solving Problems). North
Holland: Amsterdam.
Merschrod K. (1980) Partial Order Scalogram Analysis: A
Technique for Scaling Qualitative Data in Two Dimensions. Quality and Quantity,
14: 635-650.
Shye S. (1978) Partial Order Scalogram Analysis. In (Shye
S. Theory Construction and Data Analysis in the Behavioural Sciences). Jossey-Bass:
San Francisco.
Shye S. (1985) Multiple Scaling. North-Holland:
Amsterdam
Shye S. and Amar R. (1985) Partial Order Scalogram Analysis
by Base Coordinates and Lattice Mapping of the Items by Their Scalogram Roles. In
(Canter D. Facet Theory: Approaches to Social Research). Springer-Verlag: New
York.
Click here to download the zipped
program suite FACET.