You are on page 1of 12

831787

research-article2019
ASMXXX10.1177/1073191119831787AssessmentChen and Hua

Article
Assessment

Selecting Tetradic Short Forms of


1­–12
© The Author(s) 2019
Article reuse guidelines:
the Taiwan Wechsler Adult Intelligence sagepub.com/journals-permissions
DOI: 10.1177/1073191119831787
https://doi.org/10.1177/1073191119831787

Scale IV journals.sagepub.com/home/asm

Hsinyi Chen1 and Mau-Sun Hua2

Abstract
Factor-based Wechsler Adult Intelligence Scale–Fourth Edition (WAIS-IV) tetrads were investigated using the Taiwan
WAIS-IV standardization sample of 1,105 normal adults aged between 16 and 90 years. Various psychometric characteristics,
time constraints, and qualities of estimation were compared among 90 tetrads using linear equation procedures. Among
the tetrads, the Information–Visual Puzzle–Digit Span–Digit Symbol combination had higher performance than the other
combinations with respect to overall estimation quality and time saved. Moreover, the Similarities–Visual Puzzle–Digit
Span–Digit Symbol, Information–Matrix Reasoning–Digit Span–Digit Symbol, and Information–Visual Puzzle–Letter
Number Sequencing–Digit Symbol combinations obtained the most efficient estimates. For clinicians who value the utility
of Block Design, the Information–Block Design–Digit Span–Digit Symbol combination was found to provide high estimation
quality. The findings also revealed that the previously recommended Wechsler Adult Intelligence Scale–Third Edition
tetrads are no longer the best solutions for the WAIS-IV. Furthermore, even the selected WAIS-IV tetrads had substantial
misclassification rates; the four-factor short forms tend to underestimate the full-scaled IQ for highly intelligent adults.
Therefore, these short forms should be used cautiously and for screening purposes only.

Keywords
WAIS-IV, short form, Taiwan norm

The Taiwanese version of the Wechsler Adult Intelligence Taiwan, partly because of the high costs associated with
Scale–Fourth Edition (WAIS-IV) was recently released adaptation and norming. This lack of availability makes the
(Wechsler, 2015a, 2015b). Administration of the 10 development of Wechsler short forms even more important
WAIS-IV core subtests requires 79 minutes on average for in Taiwan.
normal adults (Wechsler, 2015a) and possibly longer when Prior research has examined the use of various short
working with clinical populations. Administering the full forms of the previous edition, the Taiwanese version of the
WAIS-IV battery is beneficial for maintaining the “gold Wechsler Adult Intelligence Scale–Third Edition (WAIS-
standard” (Hartman, 2009), which provides the highest pos- III; H. Chen, Hua, & Zhu, 2007; H. Chen, Hua, Zhu, &
sible reliability and validity, and the most comprehensive Chen, 2008; Tam, 2004). However, because of the structure
construct representation for evaluating intelligence. and content modifications made to the WAIS-III to form the
However, using short forms can be beneficial when full WAIS-IV (Coalson, Raiford, Saklofske, & Weiss, 2010;
testing is too lengthy and thus expensive in clinical settings Wechsler, 2008a, 2008b), it is reasonable to suspect that the
due to practical reasons or when only intelligence quotient best short forms of the two versions may differ. For exam-
(IQ) screening is required (Denney, Ringe, & Lacritz, 2015; ple, the perceptual reasoning component is strengthened in
Donders, 2001; Meyers, Zellinger, Kockler, Wagner, & the WAIS-IV, which drops two WAIS-III perceptual organi-
Miller, 2013). The need for Wechsler short forms in practi- zation subtests (Picture Arrangement and Object Assembly)
cal settings is clearly revealed by the fact that Wechsler and adds new subtests to measure visual processing (Visual
scales are the most frequently applied measures in deriving
short form IQ estimates (Thompson, Lobello, Atkinson, 1National Taiwan Normal University, Taipei, Taiwan
Chisholm, & Ryan, 2004). More important, independently 2Asia University, Taichung, Taiwan
standardized and abbreviated IQ batteries such as the
Corresponding Author:
Reynolds Intellectual Assessment Scales (Reynolds & Hsinyi Chen, Department of Special Education, National Taiwan Normal
Kamphaus, 2003, 2015) and Wechsler Abbreviated Scale of University, 162, He-Ping East Road, Section 1, Taipei, Taiwan.
Intelligence (Wechsler, 1999, 2011) are not available in Email: hsinyi@ntnu.edu.tw
2 Assessment 00(0)

Puzzles) and fluid reasoning (Figure Weights). Moreover, Method


both the Picture Completion and Arithmetic subtests, which
were core subtests in the WAIS-III, are supplemental sub- Participants
tests in the WAIS-IV (Wechsler, 2015a). With such struc- We analyzed the Taiwan WAIS-IV standardization
tural differences, updated investigations concerning the responses from 1,105 adults (548 males and 557 females)
WAIS-IV short forms are warranted. who reported no history of significant medical illness or
In the literature, Wechsler short forms vary from compris- psychiatric problems. This nationally representative sample
ing as few as two to as many as seven or eight subtests was divided into 13 age groups ranging from 16 to 90 years,
(Christensen, Girard, & Bagby, 2007; Donders, 2001; with 85 cases in each age group. The sample was carefully
Donders & Axelrod, 2002; Donnell, Pliskin, Holdnack, selected to match the 2013 Taiwan Census regarding gender,
Axelrod, & Randolph, 2007; Dumont & Faro, 1993; Girard, education level, and geographic region. A detailed descrip-
Axelrod, Patel, & Crawford, 2015; Kaufman, 1990; Kaufman, tion of the sample is provided in the technical manual
Ishikuma, & Kaufman-Packer, 1991; Ryan, Kreiner, (Wechsler, 2015b).
Gontkovsky, & Umfleet, 2015; Schopp, Herrman, Johnstone,
Callahan, & Roudebush, 2001; Silverstein, 1982, 1990;
Ward, 1990; Wechsler, 2011). Findings generally suggest that Tetrad Composition
as the number of subtests in the short form is increased, the All tetrads were required to include one subtest from each
predictive validity improves; moreover, the testing time of the four WAIS-IV factors. A total of 90 tetrads were
required also increases as a trade-off. Among the short form developed based on all possible combinations of three ver-
versions, the two-, four-, and seven-subtest combinations are bal comprehension subtests (Similarities [S], Vocabulary
the most frequently referred to. The present study focused on [V], Information [I]), five perceptual reasoning subtests
four-subtest short forms that estimate full-scaled IQ (FSIQ) (Block Design [B], Matrix Reasoning [M], Visual Puzzle
for the following reasons: (a) tetrads generally have higher [Vp], Figure Weight [F], Picture Completion [P]), three
reliability/predictive validity than dyads, (b) tetrads are time working memory subtests (Digit Span [D], Letter-Number
saving compared with short forms with five or more subtests, Sequencing [L], Arithmetic [A]), and two processing speed
and (c) tetrads include one subtest from each of the four subtests (Symbol Search [Ss], Digit Symbol-Coding [Ds]).
WAIS-IV factors—Verbal Comprehension Index, Perceptual Among all the 15 WAIS-IV subtests, only two were
Reasoning Index, Working Memory Index, and Processing excluded: Comprehension was excluded because of its high
Speed Index, which preserves the factor structure of the time requirement, scoring difficulty, and relatively larger
WAIS-IV FSIQ construct and minimizes the possibility of assessor bias variance compared with the other three verbal
overestimation or underestimation of the FSIQ for individu- comprehension subtests (H. Chen, Pan, & Zhu, 2016).
als with strengths or weaknesses in differential cognitive Cancellation was excluded because of its lowest g-loading
domains (Blyler, Gold, Iannone, & Buchanan, 2000; Donders, (.41) and relatively longer testing time compared with the
1997, 2001; Smith, McCarthy, & Anderson, 2000). other two processing speed subtests.
Three methods are frequently used for deriving short In this study, the abbreviation for each of the 90 tetrads
form FSIQ estimates: linear scaling, prorating, and regres- is the combination of the subtest abbreviations. For exam-
sion. The prorating procedure has a tendency to inflate vari- ple, a tetrad named “SBDSs” represents a combination of
ation and generate FSIQ estimates that are too extreme Similarities, Block Design, Digit Span, and Symbol
(Tellegen & Briggs, 1967) and highly sensitive to the num- Search, whereas “IPADs” is a combination of Information,
ber of subtests and specific subtests selected (H. Chen et al., Picture Completion, Arithmetic, and Digit Symbol-
2008; Engelhart, Eisenstein, Johnson, & Losonczy, 1999). Coding. Abbreviations for each of the 13 studied subtests
The regression procedure tends to yield scores with a nar- are listed in Table 1.
row range (H. Chen et al., 2007). Their estimation biases
make both the prorating and regression procedures inappro-
priate. Therefore, linear scaling methodology was selected
Procedure
for deriving FSIQ estimates because of this procedure’s To derive estimated FSIQ for each tetrad, a linear equating
well-recognized psychometric qualities (H. Chen et al., formula based on the formula proposed by Tellegen and
2007; Sattler, 2001; Sattler & Ryan, 2009; Silverstein, Briggs1 (1967, p. 504, formula 4) was used to compute a
1990; Tellegen & Briggs, 1967). deviation quotient (DQ) with a mean of 100 and a standard
The present study compared the FSIQ estimation quality deviation of 15. Under the proposed approach, short form
of various WAIS-IV factor-based tetrads. Multiple factors composite scores (sum of subtest scaled scores) and DQ are
such as psychometric properties (reliability and validity), considered equivalent if they correspond to equal standard
time efficiency, and ease of administration and scoring were score deviates (Angoff, 1984; Kolen & Brennan, 2014). This
simultaneously considered. approach is considered appropriate for estimating FSIQs
Chen and Hua 3

Table 1. Basic Characteristics and Abbreviations for Each of the 13 Studied Subtests.

Average testing
Subtest (abbreviations) g-loading rxxa r with FSIQb time (minutes)c
Verbal Comprehension Index
Similarities (S) .66 .90 .76 9
Vocabulary (V) .70 .94 .76 13
Information (I) .68 .94 .77 6
Perceptual Reasoning Index
Block Design (B) .67 .86 .72 11
Matrix Reasoning (M) .71 .88 .73 8
Visual Puzzle (Vp) .66 .87 .69 8
Figure Weight (F) .81 .91 .75 9
Picture Completion (P) .53 .83 .56 7
Working Memory Index
Digit Span (D) .72 .93 .77 9
Letter-Number Sequencing (L) .71 .90 .76 9
Arithmetic (A) .76 .90 .74 8
Processing Speed Index
Symbol Search (Ss) .56 .79 .66 3
Digit Symbol-Coding (Ds) .57 .92 .69 3

Note. FSIQ = full-scaled intelligence quotient.


aData derived from Wechsler (2015b). bData derived from Wechsler (2015b). cData derived from Wechsler (2015a).

(Sattler, Dumont, & Coalson, 2016; Sattler & Ryan, 2009) Results
and has been applied in numerous studies of the short form
literature (Girard et al., 2015; Groth-Marnat & Wright, 2016). Table 1 summarizes the basic characteristics and abbrevia-
In the present study, various types of psychometric tions of each of the 13 studied subtests. The g-loadings were
indices and base rate information were calculated for each estimated based on the confirmatory factor analysis results
tetrad to comprehensively investigate the consistency presented in Wechsler (2015b). The g-loadings ranged
between estimated and real FSIQ (Floyd, Clark, & Shadish, from .53 to .81, and were highest for Figure Weight and
2008; Kaufman, Kaufman, Balgopol, & McLean, 1996; Arithmetic. The Picture Completion, Digit Symbol, and
Ryan, Lopez, & Werth, 1998; Ryan & Ward, 1999; Symbol Search subtests corresponded with the lowest
Silverstein, 1985a, 1985b; Tellegen & Briggs, 1967; g-loadings. All subtests exhibited fair g-loadings above .50,
Thompson & Plumridge, 1999), namely (a) estimated aver- and the g loadings for almost half (46%) of the subtests
age administration time; (b) reliabilities; (c) corrected were greater than .70, indicating a good measure of g
part–whole correlations of the composite scales2; (d) paired (Lichtenberger & Kaufman, 2013). Almost all subtests had
sample t tests of the mean and variance differences of the reliabilities of more than .85, the exceptions being Picture
estimated and actual FSIQs; (e) correlations between esti- Completion and Symbol Search. The adjusted correlations
mated and actual FSIQs; (f) absolute error unit depend- with FSIQ were approximately .56 to .77 (p < .001).
ability coefficients3 (ϕ) based on generalizability theory, Vocabulary and Block Design required lengthy administra-
which was used to take into account both the consistency tion times, whereas the other subtests were relatively time
of scores and systematic differences between tetrad and efficient as expected.
full-scaled FSIQs; (g) percentage of people with the esti- Table 2 shows the reliabilities, corrected part–whole cor-
mated scores falling within the 90% and 95% confidence relations with the full battery, and administration time esti-
intervals for their estimated true FSIQ; and (h) correspon- mates for each of the 90 studied tetrads. All combinations
dence between the estimated and actual FSIQ qualitative had excellent reliabilities, ranging from .93 to .97. After cor-
categorizations based on Wechsler’s seven category crite- recting for possible inflation due to overlapping variation,
ria (2008b, p. 126, table 6.3): (a) extremely low (FSIQ 69 the correlation of each tetrad with the full battery was found
and below), (b) borderline (FSIQ 70-79), (c) low average to be promising, ranging from .89 to .93. The required
(FSIQ 80-89), (d) average (FSIQ 90-109), (e) high aver- administration time varied, ranging from 24 to 36 min. As
age (FSIQ 110-119), (f) superior (FSIQ 120-129), and (g) expected, the short forms that included Vocabulary or Block
very superior (FSIQ 130 and above). Design required much more time to administer (28-36
4 Assessment 00(0)

Table 2. Psychometric Characteristics and Estimated Time for Each of the 90 Tetrads.

Adjusted Timec Adjusted Timec Adjusted Timec


Tetrads rxx a part–whole r (minutes)
b Tetrads rxx a part–whole r (minutes)
b Tetrads rxx a part–whole r (minutes)
b

1 SBDSs .97 .92 32 31 VBDSs .95 .92 36 61 IBDSs .95 .92 29


2 SBDDs .96 .93 32 32 VBDDs .96 .92 36 62 IBDDs .96 .93 29
3 SBLSs .94 .92 32 33 VBLSs .94 .92 36 63 IBLSs .94 .92 29
4 SBLDs .95 .92 32 34 VBLDs .96 .93 36 64 IBLDs .96 .93 29
5 SBASs .94 .92 31 35 VBASs .94 .92 35 65 IBASs .94 .92 28
6 SBADs .95 .93 31 36 VBADs .96 .93 35 66 IBADs .96 .92 28
7 SMDSs .95 .92 29 37 VMDSs .95 .92 33 67 IMDSs .95 .93 26
8 SMDDs .96 .92 29 38 VMDDs .96 .92 33 68 IMDDs .96 .92 26
9 SMLSs .94 .92 29 39 VMLSs .95 .92 33 69 IMLSs .95 .92 26
10 SMLDs .95 .92 29 40 VMLDs .96 .92 33 70 IMLDs .96 .92 26
11 SMASs .94 .92 28 41 VMASs .95 .92 32 71 IMASs .95 .92 25
12 SMADs .96 .92 28 42 VMADs .96 .92 32 72 IMADs .96 .92 25
13 SVpDSs .94 .92 29 43 VVpDSs .95 .92 33 73 IVpDSs .95 .93 26
14 SVpDDs .96 .93 29 44 VVpDDs .96 .93 33 74 IVpDDs .96 .93 26
15 SVpLSs .94 .91 29 45 VVpLSs .94 .92 33 75 IVpLSs .94 .92 26
16 SV pLDs .95 .93 29 46 VVpLDs .96 .93 33 76 IVpLDs .96 .93 26
17 SVpASs .94 .92 28 47 VVpASs .94 .93 32 77 IVpASs .94 .92 25
18 SVpADs .95 .93 28 48 VVpADs .96 .93 32 78 IVpADs .96 .92 25
19 SFDSs .95 .93 30 49 VFDSs .95 .93 34 79 IFDSs .95 .93 27
20 SFDDs .96 .93 30 50 VFDDs .97 .93 34 80 IFDDs .97 .93 27
21 SFLSs .95 .93 30 51 VFLSs .95 .93 34 81 IFLSs .95 .93 27
22 SFLDs .96 .93 30 52 VFLDs .96 .93 34 82 IFLDs .96 .93 27
23 SFASs .95 .92 29 53 VFASs .95 .92 33 83 IFASs .95 .92 26
24 SFADs .96 .93 29 54 VFADs .96 .92 33 84 IFADs .97 .92 26
25 SPDSs .97 .90 28 55 VPDSs .94 .91 32 85 IPDSs .94 .91 25
26 SPDDs .95 .91 28 56 VPDDs .95 .92 32 86 IPDDs .96 .92 25
27 SPLSs .93 .89 28 57 VPLSs .94 .91 32 87 IPLSs .94 .90 25
28 SPLDs .95 .90 28 58 VPLDs .95 .91 32 88 IPLDs .95 .91 25
29 SPASs .93 .91 27 59 VPASs .94 .92 31 89 IPASs .94 .91 24
30 SPADs .95 .91 27 60 VPADs .95 .92 31 90 IPADs .95 .91 24
aComposite reliability based on Tellegan and Briggs (1967). bAdjusted part–whole correlation based on Tellegan and Briggs (1967). cEstimated by
summing corresponding subtest administration times from Wechsler (2015a).

minutes). The short forms that excluded both Vocabulary FSIQ (within approximately ±3.7 and ±4.4 IQ points of
and Block Design but included Information had the shortest the estimated true FSIQ score) and the accuracy relative to
administration time. The administration of these tetrads the correspondence between Wechsler’s seven qualitative
saved approximately 54% to 70% of the time required to intelligence categorizations.
administer the standard 10-core WAIS-IV subtests. All tetrads yielded estimated FSIQs that correlated
Table 3 presents the results for primary index of estima- highly with the actual FSIQs, with correlation coefficients
tion accuracy. The table presents various types of evidence ranging from .88 to .95 (all p < .01). The combinations that
at both the group and individual levels. The group-level evi- included Picture Completion appeared to have weaker cor-
dence comprises the mean and standard deviation of the dif- relation than the corresponding tetrads that did not include
ference between the estimated and actual FSIQs; correlations Picture Completion. The central tendency and variation of
between the estimated and actual FSIQs; paired sample t all estimated and actual FSIQ distributions matched closely.
test results for significant mean (tm) and variance (tv) differ- All 90 tetrads were found to provide reasonably accurate
ences between the actual and estimated FSIQs; and the gen- group mean FSIQ estimations; however, a few short forms
eralizability theory-based dependability coefficient (ϕ that contained both Digit Span and Symbol Search were
coefficient). The individual-level evidence comprises per- observed to result in significant score variation (p < .01).
centages of individuals with estimated FSIQ within the Under simultaneous consideration of relative differences
90% and 95% levels of confidence of the estimated true (revealed via correlations) and absolute differences
Chen and Hua 5

(revealed via mean differences), all tetrads revealed strong battery. Table 4 presents the linear equating formulae used
dependability coefficients of .85 to .95.4 Most of the tetrads for converting the short form scores into estimated FSIQs.
(87.5%) with ϕ values higher than the .90 standard when For each of the five suggested tetrads, the percentages of
making clinical decisions (Groth-Marnat & Wright, 2016; people belonging to the three score discrepancy categories
Nunnally & Bernstein, 1994). are presented in Table 5. Discrepancy was defined as the
In a comparison of estimation qualities among all tet- difference between the estimated IQ and actual FSIQ. A dif-
rads, the IVpDDs (Information, Visual Puzzle, Digit Span, ference within ±5 IQ points (close to the 95% confidence
Digit Symbol-Coding) tetrad performed more favorably interval) was considered to indicate a match; and underesti-
than all other combinations for almost all psychometric mation was defined as an estimated FSIQ of 6 or more
properties. The IVpDDs tetrad not only exhibited a high points lower than actual FSIQ; by contrast, overestimation
reliability of .96, and a high dependability coefficient of .95 was defined as an estimated FSIQ of 6 or more points higher
but also explained 90% of the actual FSIQ variance. than actual FSIQ. The percentage in each discrepancy cat-
Furthermore, 69% of the norm sample with IVpDDs FSIQ egory reveals essential base rate information.
estimates fell within the 95% confidence interval (approxi- The data in Table 5 demonstrate that for the group of
mately ±4.4 IQ points) of the estimated true FSIQ, and the individuals with FSIQ lower than 120, approximately 72%
accurate FSIQ classification hit rate was also the highest of to 87% of the estimated FSIQs were within ±5 IQ points of
all tetrads at 75%. Administering this short form required the actual FSIQs, and the percentages of overestimation and
approximately 26 minutes. Impressively, the estimation underestimation were distributed relatively evenly.
quality of IVpDDs exceeded that of all tetrads that required However, for the group with FSIQ of more than 120, the
more time (up to 36 minutes) to administer. Its optimal psy- percentages of estimations within ±5 FSIQ points were
chometric properties and shorter test duration makes lower at approximately 62% to 70%, and an imbalance was
IVpDDs the most favorable solution. observed where approximately 30% of these estimated
In addition to IVpDDs, other tetrads such as SVpDDs FSIQs were underestimated, whereas the percentage of
(Similarities, Visual Puzzle, Digit Span, Digit Symbol- overestimation was lower at approximately 5%. Moreover,
Coding), IMDDs (Information, Matrix Reasoning, Digit IMDDs appears to be the preferable tetrad when testing
Span, Digit Symbol-Coding), and IVpLDs (Information, adults with an FSIQ of lower than 79.
Visual Puzzle, Letter Number Sequencing, Digit Symbol- For each of the five selected WAIS-IV tetrads, we further
Coding) performed favorably in all psychometric and time- examined whether the mean discrepancies were signifi-
efficient domains. These three tetrads required less than 30 cantly different from zero within each ability (IQ) group.
minutes to administer and demonstrated strong reliability Figure 1 plots the mean discrepancies according to ability
and validity. Approximately 65% to 68% of the FSIQs esti- level. In the highest ability group (FSIQ ≥ 120), all five
mated using these tetrads fell within the known true 95% tetrads significantly underestimated FSIQ (t = −4.45 to
confidence interval, and their accurate FSIQ classification −7.24, p <. 01). Thus, when utilizing these factor-based tet-
hit rates were among the highest (72% to 75%). rads, underestimation tends to occur for the smartest 10% of
In the aforementioned four WAIS-IV short forms people.
(IVpDDs, SVpDDs, IMDDs, and IVpLDs), either Visual For the previous version of the Wechsler scale, Taiwan
Puzzle or Matrix Reasoning were selected to represent the WAIS-III (Wechsler, 2002), five tetrads (SMADs, SPADs,
perceptual reasoning factor. For clinicians who prefer the IPADs, SBADs, and IBADs) were discovered to result in
pure visual–motor organization information provided by the highest WAIS-III FSIQ estimating quality (H. Chen
Block Design, the IBDDs (Information, Block Design, et al., 2008). Considering some clinical practitioners and
Digit Span, Digit Symbol-Coding) is recommended; this researchers may have a special need to continue using these
required 29 minutes to administer and had high reliability WAIS-III tetrads to predict WAIS-IV FSIQ for purposes
and validity. Approximately 66% of the estimated FSIQs such as data tracking continuity, the relevant linear formu-
fell within the known true 95% confidence interval, and lae for converting the short form scores into estimated
the accurate FSIQ classification hit rate was one of the WAIS-IV FSIQs are also provided in Table 4.
highest (74%).
Collectively, the evidence indicated that the IVpDDs
Discussion
combination was superior to all the other WAIS-IV tetrads.
The IVpDDs offered one of the highest estimation accura- The purpose of this study was to evaluate the utility of fac-
cies and required only 26 minutes to administer. The tor-based WAIS-IV tetrads. At the group level, the linear
SVpDDs, IMDDs, IVpLDs, and IBDDs tetrads were also equating procedure generated distributions that favorably
favorable options. The estimation accuracies of these tet- matched the actual distributions. The majority of the gener-
rads were among the highest, as discussed, and they saved ated means and variances were close to their corresponding
64% to 67% of the time required to administer the full actual values. For short form selection, five tetrads are
6 Assessment 00(0)

Table 3. Comparison of FSIQ Estimation Quality of 90 Tetrads.

Differencea, % Within CI % Within same


Tetrads M (SD) r t(m)b t(v)c ϕd 90%e     95%f IQ category
1 SBDSs 0.05 (6.40) .91 0.27 −20.60** .89 47     51 64
2 SBDDs 0.05 (5.02) .94 0.32 1.07 .94 56     64 74
3 SBLSs 0.02 (5.39) .94 0.09 −0.15 .94 55     61 71
4 SBLDs 0.03 (5.11) .94 0.21 0.85 .94 57     65 73
5 SBASs 0.02 (6.01) .92 0.13 0.86 .92 49     56 67
6 SBADs 0.04 (5.84) .93 0.24 0.96 .93 52     58 68
7 SMDSs −0.02 (5.23) .94 −0.10 −0.03 .94 58     63 73
8 SMDDs 0.02 (5.40) .94 0.13 1.66 .94 55     62 71
9 SMLSs 0.03 (5.31) .94 0.18 −1.36 .94 54     61 71
10 SMLDs 0.03 (5.55) .93 0.19 1.56 .93 54     60 70
11 SMASs 0.00 (5.97) .92 0.02 −0.17 .92 52     60 67
12 SMADs 0.01 (6.27) .91 0.07 1.51 .91 52     57 66
13 SVpDSs −0.03 (5.29) .94 −0.18 −2.27* .94 56     63 73
14 SVpDDs 0.00 (4.88) .95 0.00 0.82 .95 58     66 75
15 SVpLSs −0.02 (5.56) .93 −0.11 0.52 .93 54     60 71
16 SVpLDs 0.01 (5.08) .94 0.04 0.65 .94 58     66 74
17 SVpASs −0.04 (6.11) .92 −0.23 0.99 .92 51     56 67
18 SVpADs −0.02 (5.78) .93 −0.14 0.67 .93 54     60 70
19 SFDSs −0.05 (5.95) .93 −0.26 2.70** .92 53     61 68
20 SFDDs 0.01 (5.70) .93 0.05 1.12 .93 52     59 70
21 SFLSs −0.01 (5.87) .92 −0.05 −0.72 .92 52     58 68
22 SFLDs 0.03 (5.85) .92 0.18 0.93 .92 50     57 68
23 SFASs −0.00 (6.78) .90 −0.01 0.95 .90 47     54 64
24 SFADs 0.01 (6.73) .90 0.05 1.03 .90 45     52 64
25 SPDSs 0.04 (7.32) .88 0.20 −20.36* .85 42     47 60
26 SPDDs 0.02 (6.50) .91 0.11 0.78 .91 47     54 66
27 SPLSs 0.00 (7.00) .89 0.02 0.18 .89 44     50 64
28 SPLDs 0.05 (6.81) .90 0.22 0.66 .90 46     52 66
29 SPASs −0.00 (7.20) .88 −0.01 −0.11 .88 42     48 63
30 SPADs 0.01 (7.12) .89 0.04 0.68 .89 42     48 62
31 VBDSs −0.02 (5.25) .94 −0.15 −3.43** .94 58     64 71
32 VBDDs 0.03 (5.02) .94 0.18 1.11 .94 59     67 74
33 VBLSs 0.03 (4.97) .94 0.21 −1.34 .94 60     67 71
34 VBLDs 0.04 (4.88) .95 0.25 1.32 .95 58     66 74
35 VBASs 0.01 (5.93) .92 0.08 1.04 .92 51     57 68
36 VBADs 0.01 (5.85) .92 0.08 0.98 .92 50     58 66
37 VMDSs −0.01 (4.95) .95 −0.07 −0.80 .95 60     66 72
38 VMDDs 0.01 (5.38) .94 0.07 1.86 .94 57     63 71
39 VMLSs −0.00 (4.85) .95 −0.02 −2.65** .95 61     67 73
40 VMLDs 0.03 (5.31) .94 0.19 2.00 .94 56     62 71
41 VMASs −0.01 (5.87) .92 −0.05 −0.13 .92 51     58 65
42 VMADs 0.00 (6.27) .91 0.02 1.65 .91 49    55 65
43 VVpDSs −0.03 (4.98) .94 −0.18 −3.20** .94 61     67 73
44 VVpDDs −0.01 (4.81) .95 −0.04 0.93 .95 59     67 74
45 VVpLSs −0.03 (5.06) .94 −0.22 −0.53 .94 60     65 70
46 VVpLDs −0.01 (4.76) .95 −0.08 0.74 .95 61     68 74
47 VVpASs −0.02 (6.00) .92 −0.12 1.29 .92 55     61 67
48 VVpADs −0.02 (5.75) .93 −0.09 0.99 .93 55     61 67
49 VFDSs −0.05 (5.79) .93 −0.28 2.52* .93 54     60 69
50 VFDDs 0.00 (5.78) .93 0.03 1.29 .93 51     59 69
51 VFLSs −0.02 (5.61) .93 −0.13 −1.87 .93 54     60 67

(continued)
Chen and Hua 7

Table 3. (continued)

Differencea, % Within CI % Within same


Tetrads M (SD) r t(m)b t(v)c ϕd 90%e     95%f IQ category
52 VFLDs 0.01 (5.71) .93 0.08 1.07 .93 51     59 67
53 VFASs −0.02 (6.77) .90 −0.08 1.01 .90 45     51 63
54 VFADs −0.00 (6.79) .90 −0.02 1.14 .90 45     52 61
55 VPDSs −0.05 (6.27) .91 −0.26 −1.27 .91 50     57 67
56 VPDDs −0.01 (6.17) .92 −0.04 0.81 .92 50     54 67
57 VPLSs −0.00 (6.38) .91 −0.02 −0.60 .91 47     54 63
58 VPLDs −0.01 (6.29) .91 −0.04 0.75 .91 48     55 67
59 VPASs −0.02 (6.88) .90 −0.09 0.14 .90 44     50 61
60 VPADs −0.00 (6.85) .90 −0.01 0.78 .90 44     49 62
61 IBDSs −0.01 (5.07) .94 −0.06 −4.31** .94 59     65 73
62 IBDDs 0.02 (4.93) .95 0.15 0.97 .95 60     66 74
63 IBLSs 0.03 (5.17) .94 0.16 −0.97 .94 56     62 71
64 IBLDs 0.02 (5.07) .94 0.15 0.79 .94 59     66 73
65 IBASs 0.02 (6.23) .91 0.09 0.92 .91 50     55 65
66 IBADs 0.03 (6.17) .92 0.15 0.91 .92 51     57 66
67 IMDSs −0.01 (4.80) .95 −0.05 −2.46* .95 62     68 72
68 IMDDs 0.02 (5.23) .94 0.13 1.74 .94 58     65 72
69 IMLSs 0.02 (5.02) .94 0.13 −2.04* .94 58     65 72
70 IMLDs 0.01 (5.43) .94 0.08 1.78 .94 56     62 71
71 IMASs 0.01 (6.14) .92 0.05 0.18 .92 50     57 66
72 IMADs 0.00 (6.51) .91 0.00 1.52 .91 48     54 65
73 IVpDSs −0.01 (4.95) .94 −0.09 −4.06** .94 59     66 73
74 IVpDDs −0.01 (4.80) .95 −0.05 0.60 .95 61     69 75
75 IVpLSs −0.01 (5.30) .94 −0.08 −0.37 .90 56     63 71
76 IVpLDs −0.02 (5.02) .94 −0.13 0.41 .94 61     68 73
77 IVpASs −0.04 (6.32) .91 −0.19 1.19 .91 51     57 65
78 IVpADs 0.01 (6.13) .92 0.03 0.72 .92 51     59 68
79 IFDSs −.05 (5.82) .93 −0.29 1.31 .93 53     60 68
80 IFDDs 0.00 (5.84) .93 0.02 1.04 .93 53     60 69
81 IFLSs 0.00 (5.90) .92 0.00 −1.26 .92 51     57 68
82 IFLDs 0.00 (5.98) .92 0.01 0.66 .92 50     57 68
83 IFASs −0.01 (7.11) .89 −0.06 1.08 .89 47     52 62
84 IFADs 0.00 (7.14) .89 0.03 0.99 .89 45     52 61
85 IPDSs −0.03 (6.30) .91 −0.18 −0.38 .91 48     54 66
86 IPDDs 0.01 (6.20) .92 0.05 0.70 .92 47     53 68
87 IPLSs 0.01 (6.64) .90 0.03 −0.24 .90 46     51 64
88 IPLDs 0.03 (6.55) .91 0.17 0.50 .91 47     53 67
89 IPASs −0.01 (7.21) .88 −0.06 0.14 .89 42     49 63
90 IPADs 0.01 (7.23) .89 0.03 0.86 .89 42     48 63

Note. FSIQ = full-scaled intelligence quotient; CI = confidence interval.


aDifference between estimated and actual FSIQ. bPaired sample student t test for mean differences. cPaired sample student t test for variance

differences. dAbsolute error unit dependability coefficient. eApproximately ±3.7 IQ points of the estimated true FSIQ. fApproximately ±4.4 IQ points
of the estimated true FSIQ.
*p < .05. **p < .01.

recommended for future usage: IVpDDs, SVpDDs, IMDDs, Practitioners are suggested to select the tetrad that most
IVpLDs, and IBDDs. On average, administering these tet- closely fits their referral question and provides useful clini-
rads requires 26 to 29 minutes, which is merely 33% to 37% cal information.
of the time required to administer the full WAIS-IV, and the In the present study, the IVpDDs corresponded with
tetrads have approximately 72% to 75% accuracy rates the highest performance for overall estimation qualities
when using the matching FSIQ categories as the criteria. among all the recommended tetrads. Both the IVpDDs and
8
Table 4. Linear Equating Formulaea for Estimating WAIS-IV FSIQ.

Age group, years

Tetrads 16-17 18-19 20-24 25-29 30-34 35-44 45-54 55-64 65-69 70-74 75-79 80-84 85-90
Selected five WAIS-IV tetrads with top qualities
IVpDDs 1.78x + 29 1.81x + 28 1.89x + 25 1.73x + 31 1.60x + 36 1.77x + 29 1.64x + 35 1.60x + 36 1.58x + 37 1.53x + 39 1.58x + 37 1.62x + 35 1.59x + 37
SVpDDs 1.80x + 28 1.82x + 27 1.93x + 23 1.70x + 32 1.64x + 34 1.83x + 27 1.61x + 36 1.58x + 37 1.58x + 37 1.53x + 39 1.58x + 37 1.55x + 38 1.56x + 38
IMDDs 1.86x + 26 1.76x + 30 1.85x + 26 1.71x + 31 1.60x + 36 1.73x + 31 1.64x + 34 1.59x + 36 1.57x + 37 1.50x + 40 1.55x + 38 1.61x + 36 1.56x + 38
IVpLDs 1.77x + 29 1.82x + 27 1.87x + 25 1.69x + 32 1.60x + 36 1.76x + 30 1.62x + 35 1.63x + 35 1.56x + 38 1.54x + 39 1.58x + 37 1.68x + 33 1.61x + 35
IBDDs 1.75x + 30 1.76x + 29 1.77x + 29 1.74x + 30 1.64x + 34 1.76x + 30 1.58x + 37 1.59x + 36 1.59x + 37 1.58x + 37 1.58x + 37 1.61x + 36 1.53x + 39
Supplemental references for researchers who have needs to use the old WAIS-III short forms in predicting WAIS-IV FSIQ onlyb
SMADs 1.87x + 25 1.68x + 33 1.93x + 23 1.64x + 35 1.58x + 37 1.70x + 32 1.59x + 37 1.57x + 37 1.54x + 39 1.49x + 40 1.56x + 37 1.59x + 37 1.62x + 35
SPADs 2.01x + 20 1.75x + 30 1.95x + 22 1.73x + 31 1.62x + 35 1.83x + 27 1.57x + 37 1.57x + 37 1.57x + 37 1.47x + 41 1.62x + 35 1.64x + 34 1.57x + 37
IPADs 1.96x + 22 1.74x + 30 1.93x + 23 1.74x + 30 1.59x + 36 1.75x + 30 1.58x + 37 1.58x + 37 1.55x + 38 1.47x + 41 1.63x + 35 1.72x + 31 1.60x + 36
SBADs 1.83x + 27 1.72x + 31 1.91x + 24 1.67x + 34 1.60x + 36 1.77x + 29 1.56x + 38 1.56x + 38 1.55x + 38 1.55x + 38 1.56x + 37 1.59x + 37 1.58x + 37
IBADs 1.75x + 30 1.69x + 32 1.79x + 29 1.67x + 33 1.57x + 37 1.70x + 32 1.55x + 38 1.58x + 37 1.54x + 38 1.57x + 37 1.57x + 37 1.65x + 34 1.58x + 37

Note. WAIS-IV = Wechsler Adult Intelligence Scale–Fourth Edition; FSIQ = full-scaled intelligence quotient.
ax = The sum of the four age-corrected subtest scaled scores. bThese WAIS-III tetrads should be used with caution because they exhibit relatively lower validity for predicting WAIS-IV FSIQ.
Chen and Hua 9

Table 5. Percentage of People in Each Discrepancy Category for the Five Selected Tetrads.

By full-scaled IQ ability level


Overall
group IQ ≤ 79 IQ (80-89) IQ (90-109) IQ (110-119) IQ ≥ 120

Tetrad Discrepancya category (N = 1,105) (n = 97) (n = 169) (n = 559) (n = 168) (n = 112)


IVpDDs Underestimates (≤−6 IQ points) 11.9 9.3 11.8 9.7 10.1 28.6
Within ±5 IQ points 77.1 78.4 77.0 78.5 79.2 66.0
Overestimates (≥+6 IQ points) 11.0 12.3 11.2 11.8 10.7 5.4
SVpDDs Underestimates (≤−6 IQ points) 13.0 17.5 10.1 9.5 11.9 33.0
Within ±5 IQ points 74.1 77.3 78.1 73.1 78.6 63.4
Overestimates (≥+6 IQ points) 12.9 5.2 11.8 17.4 9.5 3.6
IMDDs Underestimates (≤−6 IQ points) 12.8 8.3 11.2 10.6 15.5 26.8
Within ±5 IQ points 73.7 86.6 71.6 73.7 72.6 67.0
Overestimates (≥+6 IQ points) 13.5 5.1 17.2 15.7 11.9 6.2
IVpLDs Underestimates (≤−6 IQ points) 13.9 16.5 10.6 9.8 16.7 33.0
Within ±5 IQ points 74.2 74.2 72.8 77.3 73.2 62.5
Overestimates (≥+6 IQ points) 11.9 9.3 16.6 12.9 10.1 4.5
IBDDs Underestimates (≤−6 IQ points) 12.6 15.4 10.6 10.6 10.1 26.8
Within ±5 IQ points 75.0 75.3 74.6 76.0 75.6 69.6
Overestimates (≥+6 IQ points) 12.4 9.3 14.8 13.4 14.3 3.6

Note. FSIQ = full-scaled intelligence quotient.


aDiscrepancy = estimated FSIQ − actual FSIQ.

Figure 1. Mean discrepancies between estimated and actual FSIQs for each selected tetrad based on five ability levels.
Note. FSIQ = full-scaled intelligence quotient.

IBDDs were recommended by Sattler and Ryan (2009) as Donders and Axelrod (2002) outlined three criteria for
optimal tetrads for the U.S. WAIS-IV, which cross-validates short form selections: reliability higher than .90, correlation
our finding. higher than .82, and at least 81% of the estimated scores lie
10 Assessment 00(0)

within the 90% confidence interval of the full-length pre- Limitations caused by using these short forms should be
dicted score. In this study, all factor-based four-subtest short considered. The use of any short form indicates the loss of
forms met the first two criteria, but they did not meet the some information from the original full battery (Meyers et al.,
third criterion. Even with the best IVpDDs tetrad, only 61% 2013), and the ways that examinees perform on a short form
of the estimated FSIQs were within the 90% confidence and a full battery are different. Problems arise when subtests
interval (±3.7 IQ points) of the estimated true FSIQs. The are administered out of the order prescribed in the WAIS-IV
results revealed that if the prediction accuracy at the indi- manual. Examinees under shorter examination times exhibit
vidual level is the main concern, using short forms with improved concentration and motivation, use of short form
more subtests is optimal, such as the seven-subtest short examinations comprising only two subtests may result in rela-
forms (Batchelor & Meyers, 2013; Meyers et al., 2013; tively strong positive bias; however, this influence appears to
Ward, 1990). Regardless, the critical challenge of balancing be minimized for tetrads (Thompson, Howard, & Anderson,
validity and time efficiency remained. Based on current 1986; Thompson & Plumridge, 1999). In the current study,
findings, the recommended tetrads should be used for the four-subtest short forms did not have independent norms,
screening or in situations where a precise FSIQ at the indi- and the problem of overlapping error variances (Kaufman &
vidual level is not required (Sattler & Ryan, 2009). Lichtenberger, 2002) is applicable to the study findings.
Similar to findings based on the WAIS-III (H. Chen et al., Another limitation is that because we used the WAIS-IV
2008), the current results show that the FSIQ estimates gen- Taiwan standardization sample for deriving short form esti-
erated by these tetrads tended to underestimate the FSIQ for mates, findings based on the normal samples may not be fully
the top-ability group (the smartest 10% of the population). generalizable to clinical populations with various neurocogni-
One possible explanation for this findings could be that the tive deficits (J. Chen, 2017; McCusker, 1994; Reid-Arndt,
Processing Speed Index scores of members of the highest Allen, & Schopp, 2011). Further validity studies in these clini-
intelligence population tends to be lower relative to their cal populations are thus encouraged. Users are also encour-
overall intellectual profiles (Wechsler, 2008b). However, the aged to interpret short form scores with caution.
weighting of the Processing Speed Index subtests was In conclusion, these short forms can have numerous
shifted from one fifth in the WAIS-IV full battery up to one potential applications if the aforementioned limitations are
fourth in the factor-based tetrads. Changes in the cognitive- overcome. When the full battery of 10 core subtests cannot
factor weighting leads to imprecision in IQ score estimations be administered or is not required, current results suggest
(Saklofske, Gorsuch, Weiss, Zhu, & Patterson, 2005). The the utility of the IVpDDs, SVpDDs, IMDDs, IVpLDs, and
factor-based tetrads should thus be used with caution for IBDDs tetrads.
estimating FSIQs for individuals in the top-ability group.
The current results remind practitioners that the most Declaration of Conflicting Interests
favorable tetrads for representing the WAIS-III are no longer The author(s) declared no potential conflicts of interest with respect
among the best solutions for representing the WAIS-IV. In to the research, authorship, and/or publication of this article.
particular, the validity for the WAIS-III tetrads including
Picture Completion (SPADs and IPADs) was weak, with only Funding
48% of the estimated WAIS-IV FSIQs falling within the true The author(s) received no financial support for the research,
95% confidence interval, and the accurate WAIS-IV FSIQ authorship, and/or publication of this article.
classification hit rate was lower at approximately 62% to
63%. Users should be aware that the old WAIS-III tetrads Notes
exhibited relatively lower validity for predicting WAIS-IV
1. The formula used to compute the estimated FSIQ for a
FSIQ. Lower validities should be expected when using the old short form was as follows: DQ = (15 / Sc)( Xc − Mc) + 100 ,
WAIS-III tetrads for deriving WAIS-IV FSIQs, given the where Xc = sum of subtest scaled scores in the short form;
known structure and content changes in the WAIS-IV. For Sc = standard deviation of Xc; and Mc = normative mean
example, the perceptual reasoning component in the WAIS-IV of Xc.
has been strengthened by new subtests, such as Visual Puzzles 2. The formula used to compute the corrected part–whole correla-
and Figure Weights; consequently, Picture Completion, the
subtest that measures visual discrimination—the ability to ′ =
tions was as follows: rpw
∑∑r jl

distinguish between essential and nonessential details—is no 2 (n / 2) + ∑ r (t / 2) + ∑ r


jk lm

longer a core element constituting the WAIS-IV FSIQ. In the (Tellegen & Briggs, 1967, p. 501, formula 3)Where r′pw =
WAIS-IV, Picture Completion exhibited a comparatively modified coefficient of correlation between the composite
lower g-loading and lower adjusted correlation with the FSIQ. part and the composite whole; rjl = correlation between any
Therefore, we recommend that the updated WAIS-IV tetrads, subtest j included in the part and any subtest l included in
instead of the old WAIS-III tetrads, be used whenever possible the whole, where any included correlation between a subtest
for achieving the highest predictive accuracy. and itself is represented by its reliability coefficient; rjk =
Chen and Hua 11

correlation between any subtests j and k belonging to the part Coalson, D. L., Raiford, S. E., Saklofske, D. H., & Weiss, L. G.
(where subscript k is numerically larger than subscript j); rlm = (2010). WAIS-IV: Advances in assessment of intelligence. In
correlation between any subtest l and m belonging to the whole L. G. Weiss, D. H. Saklofske, D. L. Coalson, & S. G. Raiford
(where subscript m is numerically larger than subscript l); (Eds.), WAIS-IV clinical use and interpretation (pp. 3-23).
n = number of subtests included in the part, and t = number San Diego, CA: Academic Press.
of subtests included in the whole. Denney, D. A., Ringe, W. K., & Lacritz, L. H. (2015). Dyadic short
3. In this study, the formula for the dependability coefficient forms of the Wechsler Adult Intelligence Scale–IV. Archives
was as follows: ϕ = (σ2p) / (σ2p +σ2i + σ2pi,e), where σ2p of Clinical Neuropsychology, 30, 404-412. doi:10.1093/arclin/
is person universe-score variance; σ2i is constant effect for acv035
all persons due to discrepancy between short form estimated Donders, J. (1997). A short form of the WISC–III for clinical
FSIQ and real FSIQ; and σ2pi,e is interaction plus remaining use. Psychological Assessment, 9, 15-20. doi:10.1037/1040-
unmeasured error (Shavelson & Webb, 1991). This index is 3590.9.1.15
qualitatively the same as the intraclass correlation coefficient Donders, J. (2001). Using a short form of the WISC-III: Sinful
when used in a two-way model to assess absolute agreement or smart? Child Neuropsychology, 7, 99-103. doi:10.1076/
(model A.1 in McGraw & Wong, 1996). chin.7.2.99.3126
4. Results of variance component analysis also revealed that Donders, J., & Axelrod, B. N. (2002). Two-subtest estimates of
for all tetrads, the constant effect for all persons due to dis- WAIS-III factor index scores. Psychological Assessment, 14,
crepancy between short form estimated FSIQ and real FSIQ 360-364. doi:10.1037/1040-3590.14.3.360
(σ2i) was 0. This finding was logical because the linear scal- Donnell, A. J., Pliskin, N., Holdnack, J., Axelrod, B., & Randolph,
ing procedure ensured that the mean of each estimated FSIQ C. (2007). Rapidly-administered short forms of the Wechsler
was 100. Thus, no systematic difference between the means Adult Intelligence Scale–3rd edition. Archives of Clinical
of estimated FSIQ and real FSIQ. Various tetrads provided Neuropsychology, 22, 917-924. doi:10.1016/j.acn.2007.06.007
dependability coefficients as high as .95, where a ϕ value of Dumont, R., & Faro, C. (1993). A WISC-III short form for learn-
.95 indicated that 95% of the variability in scores was due ing-disabled students. Psychology in the Schools, 30, 212-219.
to the person’s true ability (σ2p = .95), 0% of the variance doi:10.1002/1520-6807
was attributed to the test battery (σ2i = .00), and merely 5% Engelhart, C. I., Eisenstein, N., Johnson, V., & Losonczy, M.
of the variance resulted from unsystematic interactions and (1999). Comparison of linear equating and prorated short forms
errors (σ2pi,e. = .05). Consequently, ϕ = (σ2p) / (σ2p + σ2i for estimating WAIS-R FSIQ in a neuropsychological popula-
+ σ2pi,e) = .95 / (.95 + .00 + .05) = .95. tion. The Clinical Neuropsychologist, 13, 95-99. doi:10.1076/
clin.13.1.95.1971
Floyd, R. G., Clark, M. H., & Shadish, W. R. (2008). The exchange-
References ability of IQs: Implications for professional psychology.
Angoff, W. H. (1984). Scales, norms, and equivalent scores. Professional Psychology: Research and Practice, 39, 414-423.
Princeton, NJ: Educational Testing Service. doi:10.1037/0735-7028.39.4.414
Batchelor, E., & Meyers, J. (2013). Estimating WAIS-IV FSIQ Girard, T. A., Axelrod, B. N., Patel, R., & Crawford, J. R. (2015).
using the Barona and Ward 7 short from. Archives of Clinical Wechsler Adult Intelligence Scale–IV dyads for estimating
Neuropsychology, 28, 547-548. doi:10.1093/arclin/act054.81 global intelligence. Assessment, 22, 441-448. doi:10.1177/
Blyler, C. R., Gold, J. M., Iannone, V, N., & Buchanan, R. W. 1073191114551551
(2000). Short form of the WAIS-III for use with patients Groth-Marnat, G., & Wright, A. J. (2016). Handbook of psycho-
with schizophrenia. Schizophrenia Research, 46, 209-215. logical assessment (6th ed.) Hoboken, NJ: John Wiley.
doi:10.1016/S0920-9964(00)00017-7 Hartman, D. E. (2009). Wechsler Adult Intelligence Scale IV
Chen, H., Hua, M., & Zhu, J. (2007). Development of the Taiwan (WAIS-IV): Return of the gold standard. Applied Neuro­
WAIS-III two subtest short forms. Psychological Testing, 54, psychology, 16, 85-87. doi:10.1080/09084280802644466
305-330. doi:10.7108/PT.200712.0305 Kaufman, A. S. (1990). Assessing adolescent and adult intelligence.
Chen, H., Hua, M., Zhu, J., & Chen, Y. (2008). Selection of factor- Boston, MA: Allyn & Bacon.
based WAIS-III tetrads in the Taiwan standardization sample: Kaufman, A. S., Ishikuma, T., & Kaufman-Packer, J. L. (1991).
A guild to clinical practice. Chinese Journal of Psychology, Amazingly short forms of the WAIS-R. Journal of Psycho­
50, 91-109. doi:10.6129/CJP.2008.5001.06 educational Assessment, 9, 4-15. doi:10.1177/073428299
Chen, H., Pan, T., & Zhu, J. (2016). It is the examinee’s IQ. 100900101
Psychological Assessment, 28, 1523-1527. doi:10.1037/ Kaufman, A. S., Kaufman, J. C., Balgopol, R., & McLean, J. E. (1996).
pas0000298 Comparison of three WISC-III short forms: Weighting psycho-
Chen, J. (2017). An exploration of the intelligence pattern in metric, clinical, and practical factors. Journal of Clinical Child
Taiwanese patients with common brain diseases and the clini- Psychology, 25, 97-105. doi:10.1207/s15374424jccp2501_11
cal utility of the premorbid estimation of intellectual function Kaufman, A. S., & Lichtenberger, E. O. (2002). Assessing adolescent
(Unpublished master thesis). National Taiwan University, and adult intelligence (2nd ed.). Boston, MA: Allyn & Bacon.
Taipei, Taiwan. Kolen, M. J., & Brennan, R. L. (2014). Test equating, scaling,
Christensen, B. K., Girard, T. A., & Bagby, R. M. (2007). Wechsler and linking: Methods and practices (3rd ed.). New York, NY:
Adult Intelligence Scale-Third edition short form for index Springer.
and IQ scores in a psychiatric population. Psychological Lichtenberger, E. O., & Kaufman, A. S. (2013). Essentials of
Assessment, 19, 236-240. doi:10.1037/1040-3590.19.2.236 WAIS-IV assessment (2nd ed.). Hoboken, NJ: John Wiley.
12 Assessment 00(0)

McCusker, P. J. (1994). Validation of Kaufman, Ishikuma, and of Clinical Psychology, 41, 676-680. doi:10.1002/1097-4679
Kaufman-Packer’s Wechsler Adult Intelligence Scale- (198509)41:5<676::AID-JCLP2270410515>3.0.CO;2-K
Revised short forms on a clinical sample. Psychological Silverstein, A. B. (1985b). Two- and four-subtest short forms of
Assessment, 6, 246-248. doi:10.1037//1040-3590.6.3.246 the WAIS–R: A closer look at validity and reliability. Journal
McGraw, K. O., & Wong, S. P. (1996). Forming inferences of Clinical Psychology, 41, 95-97. doi:10.1002/1097-
about some intraclass correlation coefficients. Psychological 4679(198501)41:1<95::AID-JCLP2270410116>3.0.
Methods, 1, 30-46. doi:10.1037/1082-989X.1.1.30 CO;2-V
Meyers, J. E., Zellinger, M. M., Kockler, T., Wagner, M., & Silverstein, A. B. (1990). Short forms of individual intelligence
Miller, R. M. (2013). A validated seven-subtest short form for tests. Psychological Assessment, 2, 3-11. doi:10.1037/1040-
the WAIS-IV. Applied Neuropsychology: Adult, 20, 249-256. 3590.2.1.3
doi:10.1080/09084282.2012.710180 Smith, G. T., McCarthy, D. M., & Anderson, K. G. (2000). On the
Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory sins of short-form development. Psychological Assessment,
(3rd ed.). New York, NY: McGraw-Hill. 12, 102-111. doi:10.1037/1040-3590.12.1.102
Reid-Arndt, S. A., Allen, B. J., & Schopp, L. (2011). Validation of Tam, W. C. (2004). The utility of seven-subtest short forms of
WAIS-III four-subtest short forms in patients with traumatic the Wechsler Adult Intelligence Scale-III in young adults.
brain injury. Applied Neuropsychology, 18, 291-297. doi:10.1 Journal of Psychoeducational Assessment, 22, 62-71.
080/09084282.2011.595456 doi:10.1177/073428290402200105
Reynolds, C. R., & Kamphaus, R. W. (2003). Reynolds Intellectual Tellegen, A., & Briggs, P. F. (1967). Old wine in new skins:
Assessment Scales. Lutz, FL: Psychological Assessment Grouping Wechsler subtests into new scales. Journal of Con­
Resources. sulting Psychology, 31, 499-506. doi:10.1037/h0024963
Reynolds, C. R., & Kamphaus, R. W. (2015). Reynolds Intellectual Thompson, A. P., Howard, D., & Anderson, J. (1986). Two- and
Assessment Scales (2nd ed.). Lutz, FL: Psychological four-subtest short forms of the WAIS–R: Validity in a psy-
Assessment Resources. chiatric sample. Canadian Journal of Behavioral Science, 18,
Ryan, J. J., Kreiner, D. S., Gontkovsky, S. T., & Umfleet, L. G. 287-293. doi:10.1037/h0079991
(2015). Classification accuracy of sequentially administered Thompson, A. P., Lobello, S. G., Atkinson, L., Chisholm, V.,
WAIS-IV short forms. Applied Neuropsychology: Adult, 22, & Ryan, J. J. (2004). Brief intelligence testing in Australia,
409-414. doi:10.1080/23279095.2014.953677 Canada, the United Kingdom, and the United Sates.
Ryan, J. J., Lopez, S. J., & Werth, T. R. (1998). Administration Professional Psychology: Research and Practice, 35, 286-
time estimates for WAIS-III subtests, scales, and short 290. doi:10.1037/0735-7028.35.3.286
forms in a clinical sample. Journal of Psychoeducational Thompson, A. P., & Plumridge, S. (1999). Two- and four-subtest
Assessment, 16, 315-323. doi:10.1177/073428299801600403 short forms of the WAIS–R: A comparative validity study
Ryan, J. J., & Ward, L. C. (1999). Validity, reliability, and stan- with a normal sample. Psychological Reports, 84, 371-380.
dard errors of measurement for two seven-subtest short forms doi:10.2466/PR0.84.2.371-380
of the Wechsler Adult Intelligence Scale–III. Psychological Ward, L. C. (1990). Prediction of verbal, performance and full
Assessment, 11, 207-211. doi:10.1037/1040-3590.11.2.207 scale IQs from seven subtests of the WAIS-R. Journal of
Saklofske, D. H., Gorsuch, R. L., Weiss, L. G., Zhu, J. J., & Patterson, Clinical Psychology, 46, 436-440. doi:10.1002/1097-4679
C. A. (2005). General ability index for the WAIS -III: Canadian (199007)46:4<436::AID-JCLP2270460411>3.0.CO;2-M
norms. Canadian Journal of Behavioural Science, 37, 44-48. Wechsler, D. (1999). Wechsler Abbreviated Scale for Intelligence
Sattler, J. M. (2001). Assessment of children: Cognitive applica- (WASI). San Antonio, TX: Pearson.
tions (4th ed.). San Diego, CA: Jerome M. Sattler. Wechsler, D. (2002). Wechsler Adult Intelligence Scale–Third
Sattler, J. M., Dumont, R., & Coalson, D. L. (2016). Assessment of chil- edition (WAIS–III): Manual for Taiwan. Taipei, Taiwan:
dren WISC-V and WPPSI-IV. San Diego, CA: Jerome M. Sattler. Chinese Behavioral Science Corporation.
Sattler, J. M., & Ryan, J. J. (2009). Assessment with the WAIS-IV. Wechsler, D. (2008a). Wechsler Adult Intelligence Scale–Fourth
San Diego, CA: Jerome M. Sattler. edition (WAIS-IV): Administration and scoring manual. San
Schopp, L. H., Herrman, T. D., Johnstone, B., Callahan, C. D., Antonio, TX: Pearson.
& Roudebush, I. S. (2001). Two abbreviated versions of the Wechsler, D. (2008b). Wechsler Adult Intelligence Scale–Fourth
Wechsler Adult Intelligence Scale-III: Validation among per- edition (WAIS-IV): Technical and interpretive manual. San
sons with traumatic brain injury. Rehabilitation Psychology, Antonio, TX: Pearson.
46, 279-287. doi:10.1037/0090-5550.46.3.279 Wechsler, D. (2011). Wechsler Abbreviated Scale for Intelligence–
Shavelson, R. J., & Webb, N. M. (1991). Generalizability theory: Second edition (WASI-II). San Antonio, TX: Pearson.
A primer. Newbury Park, CA. Sage. Wechsler, D. (2015a). Wechsler Adult Intelligence Scale–Fourth
Silverstein, A. B. (1982). Two- and four-subtest short forms of edition (WAIS–IV): Manual for Taiwan. Taipei, Taiwan:
the Wechsler Adult Intelligence Scale-Revised. Journal Chinese Behavioral Science Corporation.
of Consulting and Clinical Psychology, 50, 415-418. doi: Wechsler, D. (2015b). Wechsler Adult Intelligence Scale–Fourth
10.1037/0022-006X.50.3.415 edition (WAIS–IV): Technical and interpretative manual
Silverstein, A. B. (1985a). An appraisal of three criteria for for Taiwan. Taipei, Taiwan: Chinese Behavioral Science
­evaluating the usefulness of WAIS-R short forms. Journal Corporation.

You might also like