Pearson Clinical Assessment

19500 Bulverde Rd
San Antonio, TX, 28759
Telephone: 800 627 7271
www.PearsonAssessments.com

WMSIII to WMSIV: Rationale for Change

Since the publication of the Wechsler Memory ScaleFourth Edition (WMSIV) in January 2009, we have
received many questions regarding the changes introduced in the revision. What follows is an overview of
the changes in question and the supporting rationale that led to the construction of the WMSIV.

When we began developing WMSIV, we gathered customer inquiries and criticisms from internal and
external sources to help guide revision decisions. Although many decisions were relatively
straightforward, in some areas, expert opinions were mixed, and additional sources of information were
sought before making our decisions. As development proceeded, it became clear that the lack of
concensus meant that some of our development decisions would not be accepted equally by some
customers. We attempted to address the criticisms and complaints that our customers had with the
WMSIII and balance the resulting effects of these changes for the benefit of our customers and their
clients.

The chief customer concern with the WMSIII was related to the sensitivity of the assessment,
particularly the Faces subtest and its associated norms, to brain injury and dementia. We implemented
three changes to address this concern. One of the issues with Faces was that it appeared to be sensitive
primarily to disorders associated with social perception impairment (e.g., schizophrenia, autism,
Aspergers syndrome), which is consistent with areas of specialized processing for human faces in the
brain.Faces, as presented in the WMSIII, had some psychometric issues with relatively low reliability
(due to a high guess rate) and clinical sensitivity issues with floor problems (random responding resulted
in a low average score). We attempted to fix the psychometric and clinical sensitivity issues by making the
format more clinically sensitive. The result was a more clinically sensitive Faces subtest that had better
psychometric qualities, in general, but was still more sensitive to clinical groups with social perception
deficits. For this reason, it was not included in the core WMSIV package. The subtest can be found in
the social cognition section of the Advanced Clinical Solutions product. We understood that some customers
were satisfied with Faces as it existed in the WMSIII and apologize to those customers who were
negatively affected by the change.

The second change we made to address WMSIII sensitivity issues was to add easy items to the Verbal
Paired Associates subtest. In the WMSIII, there were floor and ceiling issues with this subtest, which
made it more difficult to identify impaired association memory. The easy items introduced in the WMS
IV allowed for a better floor across all age ranges, improving clinical sensitivity. Also, having more items
to recall allowed for better scaling of delayed recall (fewer jumps between scaled score points). For the
older adult battery, the overall differences are much smaller. In the WMSIII, there were 4 learning trials
and 8 items. The WMSIV has 4 learning trials and 10 items so the time difference should be marginal.

The third change we made was to improve the normative data by screening standardization cases for
possible cognitive impairment or suboptimal effort. Using the tools we developed for malingering
detection and the results from the Brief Cognitive Status Exam, all cases were evaluated for possible
effort or cognitive difficulties. This screening was not done on prior editions of the WMS.

Another major criticism we had from WMSIII customers was directed to the assessment of visual
memory. Many customers told us that they did not use the Family Pictures subtest because they did not
view it as a measure of visual memory. They felt Family Pictures primarily measured verbal memory, was
redundant with Logical Memory, was not culturally representative of their patient population, and was
confusing to older adults who often identified the grandfather or grandmother as the father or mother. In fact,
many customers used Visual Reproduction even though it was not a core subtest. However, these
customers did not like the scoring for the subtest in WMSIII and requested a scoring system that was
shorter and easier. During WMSIV pilot phase development, we produced a more culturally diverse
edition of the Family Pictures subtest but still saw some character confusion among older adults. This
could have been resolved with extensive exceptions within the scoring rules but that made scoring more
complex. We decided to drop the Family Pictures subtest at the pilot stage, with the understanding that
some of our customers who found the test satisfactory, would have preferred to have it remain part of
the core battery.
Given that many of our customers did not like Family Pictures and were using Visual Reproduction, we
made the decision to make Visual Reproduction a core subtest as long as we had scoring rules that, on
average, could be completed in 510 minutes. We tried multiple scoring systems, and the version that was
included in the WMSIV was reliably scored in less than 10 minutes by our scorers and had psychometric
and clinical sensitivity data comparable to longer scoring systems. We also felt that for younger subjects,
Visual Reproduction alone may not be sensitive enough to capture visual memory deficits. We needed to
develop an additional visual memory subtest.

We undertook development of a visual memory task understanding that there are no pure visual memory
tests and there will be confounding factors no matter how we approached the test (there are also no pure
verbal memory tests since all verbal memory tests assume some level of language functioning). In
reviewing the literature, it was clear that object memory (particularly, nameable objects) operates
differently than spatial memory. This is not surprising since visual information is processed in two visual
streams: a dorsal stream, which evaluates the spatial location of objects and is related to occipital-parietal
lobe functioning, and a ventral stream, which evaluates visual details and is related to occipital-temporal
lobe functioning (attaching language to a visual object). We felt that the best chance of a subtest being
visual rather than verbal would be to develop items that were hard to verbalize and required processing of
both spatial and detail elements. The immediate reaction of the examinee to the items on the Designs
subtest is that they cant do it. This is because the language centers of the brain are overwhelmed by the
stimuli; however, because there are item difficulty gradients, most subjects can get at least some points. If
the subtest had used nameable objects or simply used designs, the language centers of healthy adults
would be able to attach labels and turn the test into a visual-verbal association test. Patients with left
temporal epilepsy/lobectomy, would be put at a disadvantage compared to normal controls. In addition,
the right temporal epilepsy/lobectomy patients would be able to use language processing to facilitate
performance on this test. The purpose is not to inform the surgeon where to operate but to track the
impact of left or right surgery or injury on memory functions. Our results found that the right temporal
lobectomy group had difficulty with spatial memory but not memory for details, and the left temporal
group did well on both. In addition to providing better information about left-right differences than the
previous edition, this version of the Designs subtest is one of the more sensitive tests for indentifying
memory deficits in patients with moderate to severe TBI. The Designs subtest also correlates highly with
ratings of daily functioning after TBI. Overall, the subtest has a very good floor and ceiling, good
reliability, and good clinical sensitivity, and meets the need for a visual memory subtest for younger
subjects.

Our polling of WMSIII customers also indicated that they rarely used supplemental measures, other
than Visual Reproduction, so most of these were dropped or incorporated into the cognitive screener.
We also found that there were issues with the WMSIII Working Memory Index. The index was
comprised of one subtest and overlapped with the WAISIII Working Memory Index, making the two
indexes statistically dependent. This created problems when comparing the two indexes because the
standard significance level calculations assume the scores are independent. Also, since the WMSIII index
was made up of a visual and auditory measure, differences between the WAISIII and WMSIII indexes
could be driven by one score and not by working memory as a general construct. The decision was made
to separate the WAISIV and WMSIV indexes, where the WAISIV Working Memory Index would
include auditory working memory and the WMS-IV Working Memory Index would include visual
working memory.

Because the WMSIII had only one visual working memory measure, it was clear that at least one more
visual working memory measure would be required. In addition, we wanted to develop tasks that required
some mental manipulation of visual information. Although Spatial Span is a good measure of visual
spatial span, it does not require a lot of mental manipulation. Two tasks were developed as analogs to the
WAISIV Arithmetic and Digit Span subtests. The two subtests would also reflect the two streams of
vision noted previously. The Spatial Addition subtest has psychometric qualities very similar to
Arithmetic and correlates with academic functioning similar to Arithmetic. The Symbol Span subtest has
psychometric qualities similar to the WAISIV Digit Span subtest and correlates with academic
functioning in a similar manner. It was not expected that examiners would need to use both indexes on a
regular basis because they are highly correlated and function similarly. It was expected that examiners
would use the Visual Working Memory index only when they thought that the WAISIV Working
Memory Index was not an adequate measure of working memory due to language processing problems or
difficulties with arithmetic skills, or if there was a specific hypothesis about Visual Working Memory.

We have also heard complaints about the administration time being long. We have some suggestions to
offer in response. First, the WMSIV contains many optional procedures that are used to answer specific
questions. The recognition trials are optional and should be used only when there is a clinical question
regarding encoding versus retrieval deficits. The VPA Delayed Free Recall task is also optional and
should be used when a more general word recall measure is desired rather than an associative memory
measure. The administration of every task of every subtest of the WMSIV would be needed only for
unusual cases. Second, not using the WMS-IV Working Memory subtests can save a substantial amount
of time. The need for a second delay is caused by the need to administer the Symbol Span subtest after
Visual Reproduction plus the administration time required for Spatial Addition. When not using these
subtests, the memory tests can be administered sequentially with only a single delay.

We recognize that time and billing constraints challenge all psychologists to provide the most
comprehensive, yet cost-efficient evaluation, and we are working towards providing solutions to further
meet those needs.

We hope that this information has provided insight into the development process of the WMSIV and
the decisions and trade-offs that were made to best serve our customers and their clients. If you have
additional questions regarding the WMSIV or any of our products, you can always contact us at
800.627.7271.