Rather than continue to answer individual queries regarding
our approaches to grid data collections and analyses, I will, herewith,
send on a summary description of some of our programs.
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Computer Progams for Role Repertory Analyses
James Jaccard and James C. Mancuso
The University at Albany -- State University of New York
Twelve computer programs make up the main
body of separate suites related to the Parent Role
Repertory Analysis and Self Role Repertory Analysis.
Comparable programs have been written for Parent
Role Repertory Analysis and Self Role Repertory
Analysis. One program in each suite deals with data
collection, one program concerns data listing, two focus
on data preparation and nine programs address data an-
alysis. These twelve data analysis programs focus on
one participant (i.e., single-subject analysis). A brief
description of programs follows. Another set of
programs (available on a limited basis), described
briefly below allows for comparing sets of grids. All
programs are described in a detailed manual prepared
for use with these programs.
All of the data analysis programs have been written
in QuickBASIC and have been designed to fit the "uni-
versal data collection mode" which has been embodied
in PAREP.BAS and SELFGRID.BAS, the data collec-
tion programs described below. In order to adapt these
programs for other uses three basic changes must take
place in the easily accessed PAREP and SELFGRID
programs. (1) One must change the DATA statements
which appear at the end of the program. One can de-
scribe any general category of events or objects in the
place of the provided materials found in data statements.
(2) One can insert any judgment dimensions into the
place of the DATA statements which are written into
current programs. (3) If one wishes to vary the number
of events or objects to be used in the grids, two fixed
variables must be changed -- (a) NC, the number of
columns: columns define the events to be judged, and
(b) NR, the number of rows: rows define the judgment
dimensions along which the participant rates the events.
The manual describes the structure of the data collection
programs for users wishing to make changes.
Having carefully made these changes, one may run
all of the programs without encountering compiling,
runtime, or logical errors. One should, however, go
through all programs to adapt labels, (e. g. replace the
terms roles and beliefs with other terms in the analysis
programs, change the names of the extensions which
cue the contents of the files -- that is, replace the exten-
sion PGR, which indicates the participant's stored par-
ent repertory grid).
To allow maximum flexibility the programs can be
provided in their raw QuickBASIC format. One can
easily turn these programs into executable form, but to
do so would make them inaccessible for adaptation to
other uses.
Should a user wish to work with the programs
without altering them, he or she can acquire the
compiled versions of the programs.
QuickBASIC comes highly recommended. One can
easily learn its rudiments, and one can then proceed
easily through different levels of manipulating these pro-
grams.
Once one has a compiled program and has it
running, he or she will find it to be very user friendly.
Easily-followed instructions lead to screen displays and
to printouts.
Program 1: PAREP and SELFGRID
These programs are used to collect data from the
participant. They can accomodate up to 45 rows (con-
taining judgment dimensions) and 45 columns (con-
taining objects to be judged).
PAREP completes four general tasks with each
respondent. The first section of the program arranges
for the participant to enter a name of a known (or im-
agined parent) which he or she can associate to the role
description. The second part assesses perceptions of
each parental role vis-a-vis each belief. That is, the
participant enters a rating of her/his judgment of the
extent to which the role person would believe (ex
tremely unbelievable to extremely believable) each of
the statements about parenting. While responding to
the third section the participant provides data for
purposes of a reliability analysis. This involves re-
assessing (in a test-retest format) at least 10% (no more
than randomly chosen judgments) of the original
judgments. The fourth section elicits the participant's
judgments of overall parent success for each role.
SELFGRID works a participant through six tasks. The
participant (1) names persons with whom he/she inter-
acts; (2) names ways in which he/she might interact
with persons by entering terms which describe his/her
behaviors in social interactions, (3) rates the extent to
which he/she shows each of the named behaviors to-
ward each of the named persons, (4) repeats about 35
randomly selected judgments so that an error index can
be calculated, (5) indicates the level of comfort he or
she feels toward the names persons, and (6) indicates
the extent to which he/she wishes others to see him/her
as person who behaves in terms of the named descrip-
tors, thus providing a "centrality index" for each self-
describing dimension.
The design of the programs allow for frequent
saving of data, for at-will task interruption, and for easy
resumption of work toward completing the data col-
lection.
All data are stored on disk in a format which
allows for ready access when a user executes the
analysis programs.
Program 2: PARDAT and SELFDATA
These programs provide a printout of all of the data
collected in the PAREP and SELFGRID programs,
respectively.
Program 3: PLABELS and SELABELS
These programs are used to generate and store files
for labels for the various roles and beliefs for PAREP
(self-describing dimensions and role target persons for
SELFGRID) which are then accessed in the data analy-
tic programs. Use of these programswill eventually
facilitate interpretation of print-outs.
Program 4: PENTRY and SELENTRY
This program permits the entry of data of the type
and format one might collect by the PAREP and
SELFGRID programs. If data are collected by means
other than these data collection programs (e.g., on a
questionnaire), then this program allows the user to en-
ter the data in such a way that it can be analyzed using
all the programs described here.
Program 5: PARPRIN and SELFPRIN
These programs conduct a principal components
analysis of the matrix form data. PARPRIN analyses
can be conducted for either roles (across beliefs), or
beliefs (across roles). SELFPRIN will analyze either
self role descriptors or role target persons. Output
includes relevant means, standard deviations, correla-
tions, and a standard principal components analysis
(with a varimax rotation).
Program 6: PARCLUS and SELFCLUS
These program generates Euclidean distance scores
between beliefs (or roles) for PAREP. SELFCLUS
performs the same functions for data from SELFGRID.
These cluster analysis programs then subject these
scores to a hierarchical cluster analysis. The distance
scores are also analyzed in terms of traditional profile
analysis, decomposing each score into elevation, scatter,
and shape components. In addition, an error index of
measure reliability is computed, as well as diagnostic in-
dices for the application of either multidimensional
scaling or cluster analysis. Distance scores are stored
on a disk.
Program 7: PARSCORE and SELFSCORE
PARSCORE calculates mean belief scores for
user-specified clusters of roles or beliefs. It permits the
user to collapse across certain roles and beliefs (based
on earlier cluster analyses) and to examine mean belief
differences as a function of the grouped roles and/or
beliefs. SELFSCOR performs similar functions on
SELFGRID data, allowing the user to obtain means of
judgments within groups of dimensions and role target
persons. The groupings would be obtained through
reference to results from cluster analyses and factor
analyses.
Program 8: PARMODEL, SVALMODL, and SCEN-
MODL
The PARMODEL program analyzes the relationship
between beliefs and judgments of overall success of
each role. Two types of models are explored for each
belief. The first model is a "More-is-Better" model,
which examines the extent to which judgments of
parental success increase, in a linear fashion, as the
judgment of belief increases. The second model is an
"Ideal Point" model, in which success is predicted from
a second-order power polynomial of the belief data.
The analysis isolates the point on the belief dimension
where judged success is maximal (or minimal).
SVALMODEL performs a similar function, in that
it models each self describing dimension against the
participant's reported level of comfort in interactions
with each named role target person.
SCENMODL models each role target person
against the centrality index. In this way, the ways in
which a respondent presents his/her self to a role target
persons is contrasted to the importance, to him/her, of
presenting himself/herself in those ways.
Program 9: PARMUL and SELFMUL
PARMUL conducts a multiple regression analysis
between judged success of the various roles and a
user-specified number of beliefs. The program permits
the multivariate evaluation of the relationship between
perceived success and multiple beliefs, considered si-
multaneously.
SELFMUL will perform similar analyses of judged
comfort against self describing dimensions and
centrality indices against judgments of ways of acting
toward particular role target persons.
Program 10: PARINTER and SELFINTR
This program, like PARMUL, evaluates the
relationship between judged success of the various roles
and two beliefs, in a multiple regression context.
However, the program also permits the analysis of two
way interaction effects between beliefs, using moderated
regression methods.
Again, SELFINTR performs similar analyses of
comfort indices agains self describing dimensions and
centrality indices against ways of responding to
particular role target persons.
Program 11: PARRESID and SELFRSID
PARRESID conducts a residual analysis on corrle-
ations between a given belief and judged success (for
PAREP). The analysis takes the form of contrasting a
correlation with one case deleted as compared with the
correlation based on the full N. In this case, the N
represents the different roles. Similar analyses will re-
sult from running SELFRSID to contrast comfort
indices against dimensions or centrality indices against
judgments of ways of re-sponding to role target persons.
Program 12: PARPLOT and SELFPLOT
PARPLOT performs various plots of data, including
plots based on profile analysis of roles across beliefs,
beliefs across roles, ands scatterplots of the relationship
between beliefs and judged success.
SELFPLOT carries out comparable analyses using
either the comfort indices or the centrality indices
against dimensions or target role persons respectively.
The program does not require a graphics printer. All
output is to the screen. Output to the printer is possible
by use of the"Print Screen" key on the keyboard.
NOMOTHETIC PROGRAMS
A variety of programs facilitates the preparation of
data from grids in order to contrast the grids from two
individuals, to collapse the grids from several persons
into a single grid, to cluster analyze a series of grids,
and to contrast "average grids" obtained by collapsing
several selected grids. These programs are available,
at this time, on a limited basis.
------------------------
James C. Mancuso
Department of Psychology
University at Albany
State University of New York
1400 Washington Avenue
Albany, New York USA 12222
Bitnet: JCM61@ALBNYVMS.BITNET
Internet: JCM@UACSC1.ALBANY.EDU
Home:
15 Oakwood Place
Delmar, New York USA 12054
Telephone (518) 439-4416