RATER AGREEMENT
ON
WISC PROTOCOLS
CHARLES
K.
MILLER, NORMAN
M.
CHANSKY, AND GILBERT R. GREDLER
Temple
University
The purpose of the present study was to investigate the degree of agreement
among psychologists in training for the scoring of WISC protocols.
It
was hypothe-
sized that WISC subtests lend themselves to highly objective scoring and therefore
ratings would be highly comparable.
METHOD
All graduate students currently enrolled in course work under the School Psycholo
De artment
during the September semester at Temple Uniyersity served as subjects
(N
=
24).
adttion, all
students enrolled in
a
doctoral program in clinical psychology were includ+
(N
=
8).
All
students
(with the exception of one who was taking the course concurrently) had previously completed course
work in the administration and scoring of the WISC.
Each of the
32
trainees was given a completed WISC protocol containing the responses from six
verbal and five performance subtests (mazes eliminated). Attached to the protocol was a brief case
history
containing fictitious background
of
a school age child. The students were told to score the
test
as
they would
t
ically. The responses to the original protocol were printed and
co
ies duplicated
to
assure each ratexad
a
comparable form to score. The protocols were scored at tte rater’s con-
venience and mailed back
to
the investigators.
The responses to the various subtests were fabricated by the investigators and contained
a
sampling
of
responses not covered in Wechsler’s manual. For example, the response “Vikings” was
included under the Information item, Who discovered America? In addition several items on the
Coding subtest were deliberately recorded as incorrect responses.
RESULTS
Presented in Table
1
are the frequency distributions with their associated
probabilities for the verbal, performance and full scale IQs. Each scale reveals a
wide range of scores and consequently low rater agreement.
TABLE
1.
DISTRIBUTION
OF
SCORES
FOR
VERBAL,
PERFORMANCE
AND
FULL
SCALE
I&
Verbal
I&
Performance
I&
Full Scale
I&
Score Frequency
P
Score Frequency
P
Score Frequency
P
99
1
.031 96
1
.031
93 1 .031
97
1
.031
93 3
.093
92
1
.031
92
2
,062 92
1
.031
91 2 .062
91
1
.031
90
1
,031
88
3
.093
90
3 .093
86 3
.093
86
2
.062
89 3 .093 83 1
.031
85 1 .031
87
2
.062
82 1
.031
84
1 .031
86
5
.155
80
7
.217
83
3 .093
84 4 .124
78 1
.031
81 4 .124
82 3
.093
76
2
.062
80
7 .217
81 3 .093 75
4
.124
79 3 .093
78 1 .031
76 1 .031
85 4 .124 79 7 .217 82 2 .062
-
-
-
-
- -
-
-
- -
-
-
Table
2
presents
a
distribution of scaled scores for each of the subtests. The
picture completion subtest was omitted because the raters had difficulty interpreting
the responses on the protocol.
资源评论
