Filetype PDF | Posted on 20 Sep 2022 | 4 years ago
Authentication
digital resources
296x Filetype PDF File size 0.79 MB Source: conservancy.umn.edu
The Between Item
Relation
the
Format and the Structure of
Inventory
Eysenck Personality
F. and John F. Stevenson
Wayne Velicer
of Rhode Island
University
A seven-choice format for
Likert response per- Comrey Scales 1970) are
allows finer distinctions sub- Personality (Comrey,
inventories by
sonality one of the few to
this
trend,
jects than the traditional two-choice format. The exceptions employ-
a seven-choice item format. The
Eysenck Personality was in the ing Likert-type
Inventory employed study the value of the multi-
to test the that use of the present investigates
study
present hypothesis
based on
format would in a the that
expanded result clearer and approach,
category reasoning
more accurate indication of sub-
test structure. The such will
items the to make finer
permit subject
in a
jects, volunteers course, took the
psychology distinctions and, therefore, will provide more
standard two-choice version of the EPI and a seven- precise and This should
choice version one week with the order meaningful responses.
apart, in item and scale
result increased
counter-balanced. A analysis (1) reliability,
principal components
with a varimax two for more favorable reactions to the inven-
rotation (2)
yielded components subject
the two-choice identifiable as
format,
clearly and a clearer and more accurate indica-
tory, (3)
Eysenck’s "Neuroticism" and "Extraversion" which tion of structure.
The the test
for 18% of the variance.
accounted
together Of
seven-choice version in six ac- these questions, the reliability issue has
resulted components been the
most but
researched,
for of widely
counting 46% the variance. The conflicting
expanded
format in the of results have been For
suggested structure produced. example,
inadequacies
the defined the
EPI, factor structure more and Cook and
clearly, Deutch, (1951)
Jahoda, Ferguson
and a of the variance.
explained greater proportion (1941) report that reliability increases as the
It thus demonstrated the of
apparent advantages number of increases.
categories Bendig
the format for scale construction. response
multiple-response (1954), Komorita (1963), Peabody (1962), and
structured inven- Matell and have found
(1971)
Traditionally, personality Jacoby reliability
tories have a two-choice item format to be of the number of re-
independent
employed generally
choice Komorita and Graham
(e.g., true-false, agree-disagree, a forced sponse categories. (1965)
in
between The reasons are found an increase with an increase
two
alternatives). pri- reliability
ease of in number of for scales with rela-
(1) administration
(i.e., only
marily practical: categories
simplicity of instruction for subjects); (2) re- tively homogeneous items. Masters (1974) re-
duced ease of a relation for a scale that had an ini-
administration (3)
time; scoring; ported only
and total
avoidance of issues. The recent tial low score It is to re-
(4) scaling variation. difficult
solve these
differences because different test
(1)
were in
APPLIED PSYCHOLOGICAL MEASUREMENT instruments used the various studies and
different were
Vol. No. 2 1978 293-304 (2) methodologies employed.
2, Spring pp.
@ 1978 Co.
Copyright West Publishing
293
Downloaded from the Digital Conservancy at the University of Minnesota, http://purl.umn.edu/93227.
May be reproduced with no cost by students and faculty for academic use. Non-academic reproduction
requires payment of royalties through the Copyright Clearance Center,
http://www.copyright.com/
294
A and uncontrolled rived
potentially confounding factorially, possesses good psychometric
in-
on is the that the and
effect has been
reliability possibility properties, used The
extensively.
crease in the number of response categories limited number of items (57) permits a full
altered the structure of the instrument. The analysis at the item level. The in the
reasoning
on
structure is usually determined the two-re- preceding led to the that
paragraph expectation
form of the instrument or established the
standard EPI
with
two-choice
sponse format should
If the versions two if
three the
&dquo;theoretically.&dquo; multiple category yield (or Lie Scale
components
of the instruments have a different structure, emerges as a separate component), while a
on
then measures of internal based Likert
format
version
of the EPI
consistency should ad-
yield
in-
form be ditional
the scales of would and/or
better
(binary) defined com-
original components
if
the structure is more these should account
appropriate, particularly ponents ; components then
the Komorita and for
This would more of
the
complex. explain total variance.
Graham result with to
(1965) respect &dquo;heteroge-
control for struc-
nous&dquo; The to
scales. only study Method
ture is the recent monte carlo Lissitz
study by
an in re- The were students enrolled in two
and which found increase
Green subjects
(1975),
as the number of increased sections of a lower level psychology course.
liability categories
in- was and the students
from two to five and no for further Participation voluntary,
change
were told the nature of the but not
in number of task,
creases the categories. general
the Each
issue of reaction to of format specific student
The subject type hypothesis. completed
the standard and
has not been extensively investigated. Jones Eysenck Personality Inventory
a seven-choice version of
the EPI. The
(1968) that seven-
subjects
reported generally preferred
a form to a two-choice format. choice version was by the
multi-category developed employing
item format of
This is the first author’s in- the Comrey Personality Scales
by
finding supported
57
with the items of EPI.
formal observations. the Comrey
employed
The relation between item format and the two different scalings for different items: one
has based on a frequency concept from
structure of the not been studied (ranging
inventory
to and the other on
If in &dquo;Always&dquo; &dquo;Never&dquo;) based a
the format
extensively. multi-category does,
likelihood from to
finer concept
fact, provide meaningful distinctions, an (ranging &dquo;Definitely&dquo;
of structure at the item level could re- &dquo;Definitely Not&dquo;). This method was easily
analysis
to the items of the EPI.
in a defmed scale and struc- Administration
sult better inventory adapted
ture. in structure from was in two sessions one week
Improvements resulting apart. Approxi-
half
the students took the EPI
this would include for mately Standard
approach (1) accounting
first and then the Likert while order
more of the total (2) EPI, the
variation, higher component
and/or was reversed for the students. Only
loadings, (3) identification of additional remaining
and students who completed both forms were in-
Joe John both
(1973)
components. employed
cluded in the final =
(N 77).
forced-choice for- sample
the traditional (two-response)
For a
each version of the
and a I- inventory, principal
mat six-choice with the Rotter
response
E six-choice was on the
Scale. The format resulted in two components analysis performed
57X57 matrix of intercorrelations.
clearly factors, the item
interpretable illustrating po-
tential value of the for Velicer’s (1976) Minimum Average Partial
multi-category approach
was used to deter-
correlation method
scale structure. (MAP)
probing
In mine the number of to extract. A
the the components
study, Eysenck Personality
present
varimax was on the
was selected for use in an in- rotation performed compo-
Inventory (EPI)
of nent
of the effects format on pattern.
vestigation response
EPI A critical in of
structure. The was de- issue this is the method
inventory originally study
Downloaded from the Digital Conservancy at the University of Minnesota, http://purl.umn.edu/93227.
May be reproduced with no cost by students and faculty for academic use. Non-academic reproduction
requires payment of royalties through the Copyright Clearance Center,
http://www.copyright.com/
295
Table 1
Comparison of Results of Component Analysis with EPI Scoring Keys.*
(Continued on next page)
Downloaded from the Digital Conservancy at the University of Minnesota, http://purl.umn.edu/93227.
May be reproduced with no cost by students and faculty for academic use. Non-academic reproduction
requires payment of royalties through the Copyright Clearance Center,
http://www.copyright.com/
296
Table 1: Continued
*Loadings greater than .30 are underlined.
the number of to be an exact is
parison that
determining components purposes, stopping point
extracted. The MAP method 1976) is determined is
(Velicer, empirically Alternative,
necessary.
of advan- more
a number traditional as
such the
rather but it methods, scree test
new, possesses
this The de- or the
for a of type. procedure eigenvalue-greater-than-one criterion,
tages study
less For
termines of succes- results. both
the number by give satisfactory analyses,
components
the scree test clear
sively out until the aver- provided no solution. The
partialing components
mini-
reaches a method resulted in
age squared correlation eigenvalue-greater-than-one
partial
21 for
mum. this the and 17
retained format for
Components by procedure components binary
are clearly &dquo;common&dquo; components. For com- the Likert format. However, in both cases, the
Downloaded from the Digital Conservancy at the University of Minnesota, http://purl.umn.edu/93227.
May be reproduced with no cost by students and faculty for academic use. Non-academic reproduction
requires payment of royalties through the Copyright Clearance Center,
http://www.copyright.com/
no reviews yet
Please Login to review.
