Content uploaded by Virginia Diehl
Author content
All content in this area was uploaded by Virginia Diehl
Content may be subject to copyright.
AN INTERPRETATIVE ANALYSIS
OF
FIVE COMMONLY
USED
PROCESSING SPEED MEASURES
Gregory
M.
Feldmann
Special
Schoo/
District
of
St.
Louis
County, Town and Country,
Missouri
Ruth
M.
Kelly
and Virginia
A.
Diehl
Western lllinois University
Processing speed subtests are components of
Revised (Woodcock
&
Johnson,
1989),
and
widely used intellectual assessment insuu-
Speed of Information Processing from the
men-.
Many researchers interpret these me%
Differential Ability Scales (Elliott,
1990)
were
ures
as
assessing a unitary construct, hut there
administered
to
102
volunteer participants.
is
a
question concerning the constmcts Using regression analyses, performance on
assessed by these measures and, ultimately,
each of these tests
was
predicted by motor
their interpretative utility. Coding and
Symbol
speed and/or number facility factors.
Search from the Wechsler Intelligence
Scale
Individual differences in motor speed
were
for Children-Third Edition (Wechsler,
1991),
found to he related to each of the five process
Visual Matching and
Cross
Out from
the ing speed measures,
whereas
number facility
Woodcock Johnson
Tests
of Cognitive Ahility-
was
related to three of the measures.
This study investigated the interpretive utility of five measures of cognitive
processing speed by assessing the role of
two
skills
believed
to
underlie subtest
performance. Popular speed measures include: Coding and Symbol Search
from the Wechsler Intelligence Scale
for
Children (Wechsler, 1991, 2003);
Cross Out and Visual Matching from the Woodcock Johnson Tests of Cognitive
Ability (Woodcock &Johnson, 1989; Woodcock, McGrew,
&
Mather, 2001); and
Speed
of
Information Processing from the Differential Ability Scales (Elliott,
1990). Cognitive speed, defined as “the ability
to
fluently perform cognitive
tasks automatically, especially when under pressure to maintain focused atten-
tion and concentration” (McGrew
&
Flanagan, 1998, p.
24),
has been found to
affect working memory, fluid intelligence, and general cognitive efficiency and
influence how quickly cognitive effort can be reallocated
(Fry
&
Hale, 1996;
Kail
&
Salthouse, 1994). The five subtests
of
interest that measure processing
speed have been found
to
be positively correlated (Byrd
&
Buckhalt, 1991;
Elliott, 1990;
Kail,
1997; Wechsler, 1991; Woodcock
&
Mather, 1989) and define
a unified speed factor (Stone, 1992). When synthesizing current theories of
intelligence (Carroll, 1993; Horn, 1989), McGrew and Flanagan (1998) classi-
fied speed measures under the Broad Stratum
I1
Ability of Processing Speed
(Gs),
based on measures typically being fured-interval timed tasks requiring lit-
tle in the way of complex thinking
or
mental processing.
Even with these empirical and conceptual similarities, there is reason to
believe that process differences exist and somewhat different constructs may be
assessed. Kamphaus, Benson, Hutchinson, and Platt (1994) and Kush (1996)
questioned the theoretical and clinical utility of the WISCIII Processing Speed
Index due to its factorial structure ambiguity and the low gloadings of its con-
152
FELDMANN
ET
AL.
tributing subtests.
Also,
close inspection of Buckhalt and Jensen’s (1989)
results indicate possible differences between figural and numeric items within
the Speed of Information Processing (SIP) task. Specifically, numeric items
had smaller and fewer significant relationships with Reaction Time
(RT)
meas-
ures than did the figural items. The findings led the authors to speculate that
item content influenced performance.
Further uncertainty is raised when comparisons are made between paral-
lel
subtests on the
WISC-111
and
DAS,
such as the Vocabulary subtests,
Similarities subtests, Block Design/Pattern Construction, and Coding/SIP.
The majority of correlations between the first three parallel tasks are in the
.7
to
.8
range, whereas the processing speed tests, Coding and
SIP,
have lower cor-
relations ranging from .44 to .50 (Byrd
&
Buckhalt, 1991; Elliott, 1990; Stone,
1992). Even when corrected for attenuation, the discrepancy exists, suggesting
that the differences observed are due to something other than measurement
error and that the same construct is being assessed to a lesser degree compared
to the other parallel subtests from the Wechsler and
DAS.
Guilford’s (1967, 1982, 1988) theoretical perspective, the Structure of
Intellect Model, and the distinction made between content areas make it pos-
sible to differentiate between Coding, Symbol Search, Visual Matching,
Cross
Out, and
SIP.
According to Guilford, “Figural information is in concrete form,
as perceived
or
recalled in the form of images” (1967,
p.
227),
and “Symbolic
information is in the form of signs, materials, the elements having no signifi-
cance in and of themselves, such as letters, numbers, musical notations, and
other code elements” (1967,
p.
227).
The use of ambiguous, geometric shapes
in Symbol Search and
Cross
Out is classified as figural content, whereas the use
of digits in Visual Matching and
SIP
is classified
as
symbolic. Coding is a multi-
factor task because
it
contains both figural and symbolic elements. Kyllonen
(1993) and Neubauer and Bucik (1996) have demonstrated the importance of
task content. In this investigation, that distinction will be applied to five com-
mon processing speed measures from published intelligence tests.
There appear to he many possible skills involved
with
these
five
speed
measures, as many have speculated (Buckhalt, 1991; Buckhalt
&
Jensen, 1989;
Byrd
&
Buckhalt, 1991; Elliott, 1990; Kaufman, 1994; McGrew, 1994 Sattler,
1992; Stone, 1992). Motor speed and number facility are two
of
many con-
structs believed to play a role in differentiating individual performance on
these measures, and they were chosen for the present study.
Regarding motor speed, many labels (psychomotor skill, graphomotor
speed, paper-and-pencil
skill)
have been used, but all refer to how qnickly an
individual can write
or
copy numbers, letters, words,
or
symbols (Carroll,
1993). Each processing speed subtest has a motor component, but the degree
of motoric involvement is unclear (Baron
&
Kaye, 1984; Glosser, Butters,
&
Kaplan, 1977; Shum, McFarland,
&
Bain, 1990; Williams
&
Dykman, 1994).
Lindley, Smith, and Thomas (1988) demonstrated that paper-and-pencil-based
tasks
are influenced by motor speed. Carroll (1993) stated that psychomotor
factors are clearly distinct from measures of strict cognitive abilities, and
r
h
Pam
Fa
months
SIX^^
oi
i
ductory
course
c8
received
partlcip-
cornpen’
lnstrumc
15on
mug
a
p1
limit
was
Coding
ti
(Wechsla
ANALYSIS
OF
FIVE
PROCESSING
SPEED
MEASURES
153
attempts should he made to meamre these noncognitive skills
so
that appro-
priate adjustments to estimates of intellect can he made. At this point, there
appears to be some evidence and belief that psychomotor factors may con-
found the interpretation of processing speed measures.
According to Carroll (1993), number facility refers to skill in dealing with
numbers from counting and recognition to simple computations. Coding,
Visual Matching, and
SIP
each have a number component, and many
researchers (Buckhalt, 1991; Buckhalt
&
Jensen, 1989; Byrd
&
Buckhalt, 1991;
Kaufman, 1994; McGrew, 1994) have speculated that number facility might be
important in performance
on
these subtests, which has been somewhat sup-
ported by Stone (1992). The degree of number manipulation and quantitative
skill required
on
the three measures varies and clarification of these ambigui-
ties would be helpful.
Believing that latent variables may be responsible for variance on these
subtests, Kamphaus et al. (1994), Keith (1990, 1997), Keith and Witta (1997),
Kranzler (1997), and Riccio, Cohen, Hall, and
Ross
(1997) have called for
more research to help identify what is being assessed by the popular process-
ing speed measures. Interpretative accuracy and utility of these measures
would increase with a better understanding of specific abilities involved.
In
the
present study, it was anticipated that Coding, Visual Matching, and SIP would
have a notable numeric component (Buckhalt, 1991; Buckhalt
&
Jensen, 1989;
Byrd
&
Buckhalt, 1991; Elliott, 1990; Kaufman, 1994; McGrew, 1994; Stone
1992). It was also predicted that motor speed would be a significant predictor
of each of the five processing speed measures; it was of interest to determine
the degree to which the motor component would be predictive.
METHOD
Partkipan
ts
Participants were 102 volunteers who ranged in age from 15 years
8
months to 73 years
5
months
(M=
24 years 1 month;
SD
=
8
years 10 months).
Sixty of the participants were female and
42
were male. Sixty-five were intro-
ductory and advanced university psychology students who received extra
course credit
for
participation. Eighteen were private high school seniors who
received a small gift certificate for their participation, and the remaining 19
participants were solicited by word of mouth from the community and were not
compensated.
Instruments
Processzng
speed.
Coding requires making number/symbol associations
using a presented key. The total number of test items was 119 and the task time
limit was 120 seconds. The resulting score was the number of items correct.
Coding test/retest reliability was .79, as reported in the examiner’s manual
(Wechsler, 1991).
154
FELDMANN
ET
AL.
i.
Symbol Search requires scanning five figures to decide whether one of
two
target figures is present and then marking the appropriate
‘Yes”
or
“No”
box.
There were 45 scorable test items and a 120-second time limit. The score was
the number correct minus the number incorrect, as a correction for guessing.
Symbol Search test/retest reliability was .76 in the normative sample (Wechsler,
Visual Matching requires the identification of
two
identical numbers in a
row of six numbers. There were
60
test rows and standardized administration
allows for 180 seconds. For purposes of group administration,
150
seconds
were allowed to limit the possibility of participants finishing before standard
adminisuation time elapsed. The score was the number correct. Visual
Matching test/retest reliability was
.78
in the normative sample (Woodcock
&
Mather, 1989), although the reliability of the modified form might be some-
what different.
Cross Out requires scanning and
identifying
five figures that are identical
to
the first target figure in the row. There were 30 rows of test items and stan-
dardized administration allows for
180
seconds, but for this study 150 seconds
were allowed to limit the possibility of participants finishing before standard
adminisuation time elapsed. The score was the number of rows with all five fig-
ures correctly marked. Test/retest reliability was .74 in the normative sample
(Woodcock
&
Mather, 1989), although the reliability of the modified form
might be somewhat different.
Speed of Information Processing requires the scanning of an array of
num-
bers and circling the largest in value. There were a total
of
48 test items.
Standardized administration requires the examiner to administer six sets, eight
lines each and to record the time needed for completion of each set. For the
purpose of group administration, all
48
test items were administered during a
single administration that lasted 75 seconds. The score was the number cor-
rect. Speed
of
Information Processing testhetest reliability was
.80
in the stan-
dardization sample (Elliott, 1990), although in the modified
form
it
might be
somewhat different.
Motor
speed.
Motor speed tasks were derived from studies cited by Carroll
(1993). Construct validity evidence is provided because these measures were
found, along with other tests, to load on a factor interpreted as motor speed.
Making
Xs
requires putting an
“X”
over each lower case
0.
Rows of 10
evenly spaced
0’s
were presented
on
a single page. The score was number
of
X’s
correctly made during the 30 second time limit. The Making
Xs
test had
loadings of .47 (French, 1957) and
.80
(Scheier
&
Ferguson, 1952) on an estab-
lished motor speed factor. Test/retest reliability was .92 (French, 1957).
Writing Digits requires writing the digits “123” on rows of evenly spaced
lines. The score was number
of
times 123 was completely written within the 45-
second time limit. Validity for Writing Digits type tasks was provided by loadings
on a motor speed factor of .68 (French, 1957) and .58 (Scheier
&
Ferguson,
1952). Test/retest reliability was .91 (French, 1957).
Writing Letters requires writing the word “lack” on
rows
of evenly spaced
1991).
2%
3
4
5
6
7
8
9
10
11
Note
-:
-~
Proce&
A
C*
trained
m
IstraUon
,
ANALYSIS
OF
FIVE PROCESSING SPEED MEASURES
155
lines. The score was the number of times lack was completelywitten within the
45-second time limit. Writing Letters
type
tasks had loadings
of
50
(French,
1957),
.64
(Pemberton, 1952), and
.55
(Scheier
&
Ferguson, 1952) on a motor
speed factor. Test/retest reliability was .92 (French, 1957).
Numberjanlity.
Number facility tasks were also cited by Carroll (1993). The
tests were taken from the Educational Testing Service
Kit
of
Factor-Referenced
Cognitive Tests (Ekstrom, French, Harman,
&
Dirmen, 1976).
Addition requires the summing of sets of three one-
or
twodigit numbers.
One hundred twenty seconds were allowed to answer as many as
60
problems.
The
score
was the number correct. Basic addition tests have been
shown
to be
related to an established number facility factor with the following loadings:
.72
(Coombs, 1941), .67 (Kelley, 1964), and .63 (Roff, 1952). Test/retest reliabili-
ty
was .93 (Ekstrom et al., 1976).
Division requires the division of a
two-
or threedigit number by a single-
digit number. One hundred twenty seconds were allowed and the score was the
number correct. Division tests as a measure of number facility were observed
with weightings
on
a number facility factor. Observed loadings were
.70
(Christal, 1958),
56
(Fleishman
&
Hempel, 1954), and
67
(Kelley, 1964).
Test/retest reliability was .94 (Ekstrom et al., 1976).
Subtraction and Multiplication allows participants 120 seconds for answer-
ing rows that alternate subtraction and multiplication problems. The score was
the number correct. Christal (1958), Coombs (1941), Ekstrom, French, and
Harman (1979), and Kelley (1964) showed that Subtraction and Multiplication
tests were measures
of
number facility, with factor weightings of .70, .64,
54,
and
67,
respectively. Test/retest reliability
was
.92 (Ekstrom et al., 1976).
Table
1
Administration Order and Number
of
Parficipants Completing
Each
Form
(in parentheses)
Form
Order
1
-
2
3
4
5
6
7
8
9
10
A
(25)
Coding
Addition
Symbol Search
Making
X's
Visual Matching
Division
Writing Digits
Cross Out
Sub./Multi.
SIP
B
(26)
Visual Matching
Making
Xs
Symbol Search
Addition
Coding
Writing Letters
SIP
Sub./Multi.
Cross Out
Writint. Dieits
C
(26)
Division
Writing Digits
Cross Out
Sub./Multi.
SIP
Writing Letters
Coding
Addition
Symbol Search
Makine
X's
D
125)
Writing Letters
SIP
Sub./Multi.
Cross Out
Writing Digits
Division
Visual Matching
Making
Xs
Symbol Search
Addition
I"
I
11
Writing Letters Division Visual Matching Coding
Note.-Sub./Multi.
=
Subtraction and Multiplication;
SIP
=
Speed
of
Information Processing.
Procedure
A certified school psychologist
or
school psychology graduate student
trained in standardized administration procedures completed the test admin-
istration in a single session. Four forms were used to minimize the confound-
FELDMANN
ET
AL.
156
ing of individual differences with order. See Table
1
for test order and number
of participants who completed each form.
Standardized administration procedures were employed, with the excep-
tion
of
group administration and specific changes noted in the Instruments
description. Sample items were demonstrated on an overhead projector during
the instruction phase of each test, and scoring was completed as outlined. Two
trials of each factor marker test were completed and averaged
as
recommend-
ed by the test authors.
Table
2
Correlation Matrix (Including Means and
Standard
Deviations)
Variable
1
23 4
5
6
7
8
9 1011 12
I
ke((m0nths)
-
2
Ging
-.39**
-
38mbol
Search
-.52** .51**
-
~r
4
Visual
Matching
-.30* .62*'
.a**
-
8Crossout
-.39" .53** .57** .61**
-
6
SIP
-.03
29' .43** .84" .45"
-
7
Makingx's
-.38**
.82**
SO**
SO**
.M*'
.36**
-
8WtingDigii
-.17
.38**
.M**
.Ma*
.37** .27' .65**
-
9WtingLmels
-.28' .38** .34" .42** .35'* .22 .63** .7S'
-
1OAdditicNl
24 .13 .19 .38**
.16
.41** .15 .I9
.07
-
11
Division
.03 .15 .22 .27*
.I4
.27*
.08
07
-.01
.67"
-
12
Sub/Multi.
.ll .22 .26' .39**
.I3
.35** .24 .29*
.19
.76*'
.71**
-
M
289.9 73.1 37.5 26.0 80.4 37.2 81.8 36.3 28.3
17.1
11.4 24.7
u)
105.7 13.2
8.9
3.4 5.2 4.8
8.0
4.4 4.1 4.8 6.1
8.2
Note.-SubJMulti.
=
Subtraction and Multiplication;
SIP
=
Speed
of Information Processing.
*p
<
.01.
"p
<
,001
RESULTS
Raw scores were used during analyses because they allowed direct com-
parison across subtests, given that standard scores vary by scale and could not
always be obtained because
the
age of some participants extended beyond
established norms.
A
correlation matrix, means, and standard deviations are
presented in Table
2.
Because the magnitude of the relationship between any
two meamres is limited by the reliability of each, the obtained correlations
were corrected using a formula from Nunnally
(1978).
Disattenuated correla-
tion coefficients are presented in Table
3.
Scatter plots
for
each correlation
indicated a linear relationship between the variables.
A
comparison of means
across
the
four forms showed that Form
3
participants performed better than
Form
4
participants on all three motor speed measures, and better than Form
1
participants on one motor speed measure. This effect is best explained by the
mean age of Form
4
participants being greatest and by the negative correlation
between motor speed and age (see Table
2).
No
other significant differences
between forms were found.
Wi
w
are
I
tor
~
indi
mu,
Table
I
Factor
I
Varr:'
Addition
DIVISIM
subbat;,
Making
X
WrIhng
D
Writing
I
%
01
vact
-
-
-
__
The
;
lnvolvlng
defined
i
speed
an
soning
o
qtated
rh
refel
in
dc
obJe(
advai
ing
n
ANALYSIS OF FIVE PROCESSING SPEED MEASURES
157
Table
3
Disattenuated Correlation Matrix
Variable
12
-i
4
5
6
7
R
9
in
2SymbolSearch
.66
-
3
Vlsual Matching
.79 .78
-
4
Cross
Out
.69 .76
.80
-
5
SIP
.36
.55
.68 .58
-
6
Making
X’s
.61 .60 59 .53 .42
-
7
Writing Digits
.45
33
52 .45 32 .71
-
8
Writing Letters
.45 .41
.50
.42 .26 .68 .82
-
9
Addition
.15
.23 .45 .19 .48 .16 .21 .08
-
10
Division
.17 .26 .32 .17 .31 .09 .76
-.01
.72
-
11
SubMulti.
.26 .31 .46 .16 .41 .26
.32
.21 .82 .76
Note.-SubJMulti.
=
Subtraction and Multiplication;
SIP
=
Speed
of Information Processing.
Principal components extraction using Varimax rotation was performed
with SPSS Factor Analysis on the six factor marker tests. Loading criteria of
>.4
was used; all six variables loaded on one of
two
factors. Actual factor loadings
are shown in Table
4.
Examination of these results indicates extremely high fac-
tor
loadings (where predicted) and clearly defined factors. Communalities
indicated that the variables were well defined by this solution; the weakest com-
munality for variables from factors was
.72.
Table
4
Factor Loadings from Principal Component Analyis
Using
Varimax Rotation
Variable
1 2
Addition
.90 .08
Division
39 -.04
SubtractiordMultiplication
.90 .21
Making
X’r
.10 .84
Writing Digits
.13 .90
Writing Letten
.oo
.90
Factor
.
%of
Variance
47.15 33.01
The first factor was termed Number Facility because the three measures
involving basic quantitative
skill
had high loadings. Ekstrom et al. (1976)
defined this factor
as
“The ability
to
perform basic arithmetic operations with
speed and accuracy. This factor is not a major component in mathematical rea-
soning
or
higher mathematical skills” (p.115). Similarly, Carroll (1993) has
stated that number facility
refers simply
to
the degree
to
which the individual has developed skills
in dealing with numbers, from the most elementary skills of counting
objects and recognizing written numbers and their order,
to
the more
advanced skills of correctly adding, subtracting, multiplying, and divid-
ing numbers (p.
469).
158 FELDMANN
ETAL.
The second factor was labeled Motor Speed. Tasks that require partici-
pants to perform numerous, simple pencil motions with minimal cognitive
demands in a short amount
of
time defined this factor. The combined variance
explained by the
two
factors was 80.16%.
The results of a factor analysis with
an
oblique rotation were almost iden-
tical to that with an orthogonal rotation. This finding suggests that the solution
is stable, and confidence can be placed in the factors obtained and being used
in additional analyses.
The
two
factors obtained using the orthogonal (varimax) rotation were
used as the independent variables in the stepwise multiple regression to pre-
dict performance on each of the five processing speed measures. Tolerances
of
greater than
.90
indicated an absence of multicollinearity and singularity, and
the regression assumptions (normality, linearity, and homoscedasticity
of
resid-
uals) were found to be met. A significance level
of
p
<
.01
was used to protect
against finding significant values by chance. Age was forced to enter first, to
stab
tistically control for differences in age. Results are shown in Table
5.
Table
5
Stepwise Regression Analyses
Using
Factors
to
fredict
Petformanre
on
Pmcessing
Speed
Measures
Processing
Cndine
Aee
.I5
l1.1001= 18.15 -.04 .31
.oo
Speed
Measure Factor
R2
F
B
P
P
~"
-~~
Mitor
Speed
.I3
(2,991= 19.58 4.88
.37
.oo
Symbol Search
Age
.27 (1,1001
=
36.39 -.03 -.47
.oo
Motor
Speed
.10 (2,99)
=
28.27 1.84
.31
.oo
NumbeiFacility
.08 (3.98)
=
26.10 1.71 .29
.oo
Visual Matching Age
.09 (1,100)
=
9.73
-.01
-.23 .01
MotorSpeed
.I7
(2,991=17.04 2.11
.41
.oo
Number Facility
.14 (3,981
=
21
.SO
1.97 .38
.oo
Cross
Out
Age
.I5
(1,100)= 17.97 -.01 -.32
.oo
Motor
Speed
.10
(2,99)=
16.42
1.09
.32
.oo
Number Facility .I4 (2,99)
=
8.08
1.65 .3G
.oo
Motor
Speed
.07 (3,981
=
8.90 1.30 .29
.oo
SIP
Age
.oo
(1,100)=.12
-.oo
-.01 .9G
Age accounted
for
a significant proportion
of
the variance in all but one
of
the processing speed measures (i.e.,
SIP)
when forced to enter first. Motor
Speed accounted for significant amounts ofvariance on each
of
the processing
speed measures. The degree of involvement varied
for
each measure: Visual
Matching with
R2
=
.17,
Coding with
R2
=
.13,
Symbol Search with
R2
=
.lo,
Cross Out with
R2
=
.lo,
and
SIP
with
R2
=
.07.
The second predictor variable,
Number Facility, was found to be a component of three suhtests: Visual
Matching with
R2
=
.14,
Symbol Search with
R2
=
.08,
and
SIP
with
R2
=
.07.
No
significant relationship was found between Number Facility and the remaining
two
dependent variables.
ANALYSIS
OF
FIVE
PROCESSING
SPEED
MEASURES
159
DISCUSSION
This investigation into specific skills that contribute to individual differ-
ences on five common measures of processing speed discovered some inter-
esting relationships. First, Motor Speed was able to account for varied
(7%
to
17%) but significant amounts of variance on
all
of the speed subtests. These
results are interesting but not surprising, given that each task requires at least
some amount of paper-and-pencil involvement. Even Symbol Search and
SIP,
whose motor component appears negligible, were found to vary as a function
of motor skill. Carroll (1993) and Lindley et al. (1988) have stated that peri-
pheral motor demands may cause individuals to perform differently on these
types of measures, and this was supported by the data.
Also,
Carroll (1993) has
stated that psychomotor factors are clearly distinct from measures
of
strict cog-
nitive abilities, and that attempts should be made to measure these noncogni-
tive skills
so
that appropriate adjustments to estimates of intellect can be made.
The finding that paper-and-pencil skill had a significant role in all five pro-
cessing speed subtests
will
be helpful to practitioners as they work to accurate-
ly interpret test profiles and clearly explain variability in an individual’s skills.
The role of motor skill during manipulative, object-based tasks is becoming
more of an issue in contemporary ability assessment, because newer instru-
ments are increasingly emphasizing the ‘‘level’’
or
accuracy of performance,
rather than speed of responses, and some instruments now have the means to
account for differences in motor skill when considering an individual’s profile
(e.g., the Coding copy option).
Also of interest was the discovery that the content (numeric vs. figural) of
a task generally influenced how an individual performed, with the surprising
exception of Symbol Search. This conclusion was reached from the expected
finding that Number Facility was found to account for a significant amount of
the variance in Visual Matching and
SIP,
two
subtests whose stimuli were com-
pletely numeric. In contrast, Coding and Cross Out, whose item content was
either mixed
or
entirely figural, were not found to vary as a function of indi-
vidual strengths
or
weaknesses in Number Facility. The idea that
task
content
may influence individual performance has been suspected by many (Buckhalt,
1991; Buckhalt
&
Jensen,
1989; Byrd
&
Buckhalt, 1991; Kaufman, 1994;
McCrew, 1994; Stone, 1992). Although all five processing speed subtests could
accurately be classified under the common label of perceptual speed, it
appears, as French (1976) and Carroll (1993) have stated, that perceptual
speed may be a “centroid”
of
more specific and narrow factors. That is, per-
ceptual speed itself is an ability that can be broken down into more specific
components, based on item content. Practitioners can benefit from this find-
ing as they interpret test profiles. The anomalous finding with Symbol Search
is difficult to explain, although a latent, presently unidentified variable might
be involved.
Although participant mean age was greater than the intended audience
for three of the five processing speed subtests used in this study, and previous
~
~
~
I
160
FELDMANN
ET
AL.
editions
of
newer instruments were utilized in this study, obtained results
should be generalizable to the adult population and newer instruments. Age
range is not
an
issue for the Cross Out and Visual Matching, because these sub-
tests can be used up to
92
years of age.
An
item-by-item comparison indicates
that
WJIII
Visual Matching is identical in demands to WJR Visual Matching.
Although Cross Out has been moved to the Diagnostic Supplement,
task
demands remain essentially unchanged.
The Wechsler speed subtests that exist on both the child and adult version
(Wechsler,
1997)
are highly similar in regards to the nature
of
stimuli, number
of items, and time constraints. Additionally, these parallel subtests are used sim-
ilarly to define a processing speed factor. A comparison between
WISGIII
Coding and
WISC-IV
Coding shows them to be identical;
Symbol
Search items
also remain unchanged, although more total items are now possible on the
new version.
While
the composition of the other three factors has changed, the
speed subtests and factor structure remain essentially the same.
SIP
is intend-
ed for individuals younger than
18
years of age, but no participants finished all
of
the items,
so
individual differences were still identified and the obtained
results should be valid. Although revised batteries are now available, the
demands of the processing speed tests remain either unchanged from earlier
versions or highly similar, lending the present conclusions applicable to the
adult version and new editions.
Although interesting and statistically significant results were found, certain
limitations exist. One limitation of this investigation involved having strayed
somewhat from standardized administration, specifically, group testing, short-
ened time limits on Visual Matching and Cross Out to minimize ceilings, and
a minor alteration to
SIP
presentation. Even with the subtest modifications and
a small departure from standardized administration,
it
is likely that the tests
maintained their integrity and continued to assess the same constructs,
because item content remained unchanged.
Although the regression analyses were able to account for significant
amounts of well-defined variance on the five processing speed subtests, the
majority
of
the subtests’ variability remained unexplained. Additional predic-
tor variables may have been helpful.
A
few
of
the many
skills
that may be relat-
ed to performance
on
these measures include visual memory, attention, work-
ing memory, and visual scanning, yet, as Buckhalt (personal communication,
October
2,2000)
has stated, individuals tend to coordinate many subsystems in
performing all tasks, even ones that
on
the surface appear to be “simple,”
so
confidently
identifying
specific subskills involved may be difficult. Also, it
would be of interest to determine which processing speed subtests are most
related to the most basic of speed measures, such as Jensen-like Inspection,
Reaction, or Movement Time measures, as well as how well each of the meac
ures relates to a general intellectual measure or acquisition of specific aca-
demic
skills.
Also, because motor skill was found to play a role in the process-
ing speed measures,
it
would be worthwhile to determine if, and
to
what
degree, noncognitive skills (e.g., dexterity, motor
skill)
are involved in other
4
Q
0
El.’
ANALYSIS
OF
FIVE PROCESSING
SPEED
MEASURES
161
subtests from
common
batteries. Additionally, clarifying
why
Symbol Search
was predicted
by
the Number Facility factor in the present study would
be
enlightening.
The
goal
of
this investigation
was
to increase the interpretive utility
of
five
common processing speed subtests. Although
the
results are not exhaustive,
they
should prove helpful to individuals who use these instruments
in
research
or applied settings. Specifically,
a
better understanding of skills involved in
these types of measures should prove helpful in the development and refine-
ment
of future measures.
Also,
practitioners
will
benefit from
the
increased
interpretative accuracy of
these
processing speed subtests
during
their evalua-
tion process.
REFERENCES
Baron,
M.
B.,
&
bye,
D.
B.
(1984).
A
validity study of the WISGR Coding
B
subtest.
Journal
of
Psychoeducational
Assessment,
2,
191-197.
Buckhalt, J.
A.
(1991).
Reaction time
measures of processing speed Are they
yielding new information about intelli-
gence?
Personality and Individual
Diferences,
12,
683-688.
Buckhalt, J.
A,,
&
Jensen,
A.
R.
(1989).
The British Ability Scale Speed of
Information Processing subtest: What
does
it
measure?
British Journal
o/
Educational Psychology,
59,
100-107.
Byrd,
D.
P.,
&
Buckhalt, J.
A.
(1991).
A
multitrait-multimethod construct
validity study
of
the Differential Ability
Scales.
Journal of Psychoeducational
Assessment,
9,
121-129.
Carroll, J. B.
(1993).
Human cognitive
abili-
ties:
A
suruqr offactor analytic studies.
New York Cambridge University Press.
Christal. R.
E.
(1958).
Factor analytic
study
of
visual memory.
Psychological
Monographs:
General
and Applied,
72
(13, Whole No.
466).
Coombs, C. H.
(1941).
Afactmial
study
oj
number ability. Psychometrih,
6,
161-189.
Ekstrom, R. B., French,
J.
W.,
&
Harman,
H. H.
(1979).
Cognitive factors: Their
identification and replication.
Multivariate Behavioral Research
I
I
~
Ekstrom,
R. B.,
French,
J.
W.,
Harman, H.
H.,
&
Dirmen,
D.
(1976).
Manual for kit
of
factor-referenced cognitive tests
(3rd
ed.). Princeton, NJ: Educational
Testing Service.
Elliott, C. D.
(1990).
Differential
Ability
Scales:
Introductory and technical hand-
book.
San Antonio, TX: The
Psychological Corporation.
Fleishman,
E.
A.,
&
Hempel,
W.
E.
(1954).
Changes in factor strncture of a com-
plex psychomotor test
as
a function of
practice.
Psychometrika, 19,
239-251.
French, J.
W.
(1957).
The factorial invari-
ance
of
pure-factor tests.
The
Juurnal
of
Educational Psychology, 48,
93-109.
Fry,
A.
F.,
&
Hale,
S.
(1996).
Processing
speed, working memory, and fluid
intelligence: Evidence
of
a develop
mental cascade.
Psychological Science,
7,
237-241.
Glosser, G., Butters,
N.,
&
Kaplan,
E.
(1977).
Visuospatial processes in brain
damaged patients on the Digit Symbol
Substitution Test.
International Journal
ofNmroscience,
7,
59-66.
Guilford,
J.
P.
(1967).
The nature of human
intelligence.
New York McGraw-Hill.
Guilford,
J.
P.
(1982).
Cognitive psychole
gy’s
ambiguities: Some suggested
remedies.
Psychological
Revirw,
89,
48-
59.
Guilford, J. P.
(1988).
Some changes in
the Structure of Intellect Model.
Educational and
Psychologzcal
Measurement, 48,
14.
FELDMANN
ET
AL.
162
Horn,
J.
L.
(1989).
Models of intelligence.
In
R.
L. Linn (Ed.),
Intelligence:
Measurement, theory, and
public
poliq
(pp.
29-73).
Urhana, 1L University of
Illinois Press.
Kail,
R.
(1997).
Phonological skill and
articulation time independently con-
tribute
to
the development
of
memory
span.
Journal
of
Experimental Child
Psychology, 67,
57-68.
Kail,
R.,
&
Salthouse, T.
A.
(1994).
Processing speed as mental capacity.
ACTA Psychohgica,
86,
199-225.
Kamphaus,
R.
W., Benson, J., Hutchinson,
S.,
&
Platt,
I..
0.
(1994).
Identification
of factor models for the WISC-111.
Educational and Psychological Measure-
mnt,
54,174186.
Kaufman, A.
S.
(1994).
Zntelligmt testing
wlth the
WISGIII.
New York Wiley.
Keith, T.
Z.
(1990).
Confirmatory and
hierarchical confirmatory analysis
of
the Differential Ability Scales.
Journal
of
Psychoeducational Assessment,
8, 391-
405.
Keith,
T.
Z.
(1997).
What does the WISC-
I11 measure?
A
reply
to
Carroll and
Kranzler.
School Psychology
@arterly,
12,
117-118.
Keith, T.
Z.,
&
Witta,
E.
L.
(1997).
Hierarchical and crossage confirmato-
ry
factor analysis
of
the
WISC-111: What
does it measure?
School Psychology
~~
@avterly,
12, 89-117.
Kellev. H. P.
f1964).
Memorv abilities:
A
,.
factor analysis.
Psychometric Monographs.
(Psychometric Society No.
11).
Richmond,
VA
William ByTd Press.
Kranzler, J. H.
(1997).
What does the
WISGIII measure? Comments on the
relationship hetween intelligence,
working memory capacity and infor-
mation processing speed and efficien-
cy.
School Psycholngy Quarterly,
12,
11@
116.
Kush,
J.
C.
(1996).
Factor structure of the
WISGIII for students with learning dis-
abilities.
Journal
of
Psychoeducational
Assessment, 14,
32-40.
Kyllonen,
P.
C.
(1993).
Aptitude testing
inspired by information processing:
A
test
of the four-source model.
The
Journal of General Psychology,
120,
375-
405.
Lindley,
R.
H.,
Smith, W.
R.,
&
Thomas, T.
J.
(1988).
The relationship between
speed of information processing as
measured hy timed paper and pencil
tests and psychometric intelligence.
Intelligence, 12,
17-25.
McGrew,
K
S.
(1994).
Clinical interpreta-
tion
of
the Woodcockfohnson Tests of
Cognitive Abilify-Revised.
Boston:
Allyn
&
Baron.
McCrew,
K
S.,
&
Flauagan,
D.
P.
(1998).
The Intelligence Test Desk Reference
[n'DR).
Needham,
MA:
Allyn
&
Bacon.
Neuhauer,
A.
C.,
&
Bucik,
V.
(1996).
The
mental speed-IQ relationship: Unitary
or
modular?
Intelligence,
22,
23-48.
Nunnally, J.C.
(1978).
Psychometric theory
(2nd ed.). New York McGraw-Hill.
Pemherton, C.
(1952).
The closure factors
related
to
other cognitive processes.
Psychometrika, 17,
267-288.
Riccio, C.
A,,
Cohen,
M.
J.,
Hall, J.,
&Ross,
C.
M.
(1997).
The third and fourth fac-
tors
of
the
WISGIII: What they don't
measure.
Journal
of
Psychoeducational
Assessment,
15,
27-39.
Roff, M.
E.
(1952).
A
factorial study of the
tests in the perceptual areas.
Psych@
metric Monographs.
(Psycometric Society
No.
8).
Colorado Springs, Co.: Denton
Printing Company.
Sattler,
J.
M.
(1992).
Assessment ofchildren:
Revised and updated
(3rd ed.). San
Diego,
c1:
Jerome
M.
Sattler,
Publisher.
Scheier,
I.
H.,
&
Ferguson,
G.
A.
(1952).
Further factorial studies
of
tests of
rigidity.
Canadian Journal of Psychology,
6,
18-30.
Shum, D. H.
K,
McFarland,
K
A,,
&
Bain,
J.
D. (1990).
Construct validity of eight
tests
of
attention: Comparison
of
nor-
mal and closed head injured samples.
The Clinical Neurvpsychohgist, 4,
151-
162.
Stone, B. J.
(1992).
Joint confirmatoryfac-
tor analyses of the DAS and WISC-R.
Journal ofSchool Psychology,
30,
185-195.
Wechsler, D.
(1991).
Wechsler Intelligence
Scale
for
Children-Third Edition.
San
I
ANALYSIS OF FIVE PROCESSING SPEED MEASURES
163
Antonio, TX: The Psychological
Corporation.
Wechsler,
D.
(1997).
Wechsler Adult
Intelligence Scak-Third Edition.
San
Antonio, TX The Psychological
Corporation.
Wechsler,
D.
(2003).
Wechcler Intelligence
Scale
fo7
Childrm-Fourth Edition
San
Antonio, TX The Psychological
Corporation.
Williams, J.,
&
Dykman,
R.
A.
(1994).
Nonverbal factors derived from chil-
dren's performance on neuropsych-
logical test instruments.
DeveQmental
NeuropsYchology,
lO,19-26.
Woodcock,
R.
W.,
&
Johnson, M.
B.
(1
989).
Woodcock-Johnson Psycho-
Educational Battery-RNised.
Chicago:
Riverside.
Woodcock,
R.
W.,
&
Mather, N.
(1989).
WJ-R
Tests
of
Cognitive Ability-
Standard and Supplemental Batteries:
Examiner's manual.
In
R.
W.
Woodcock
&
M.
B.
Johnson,
Woodcock-
Johnson Psycho-Educational
Battery-
Revised.
Chicago: Riverside.
Woodcock,
R.
W., McGrew,
K.
S.,
&
Mather, N.
(2001).
Woodcock-Johnsun
III
Tests
of
Cognitive
Abilities. Itasca,
IL.
Riverside.
