ASSIGNMENT 2 – TRAUMATIC BRAIN INJURY

ANNOTATED LECTURE NOTES

 

What follows is a written version of the material presented in lectures and the residential schools designed to achieve this goal of aiding you in mastering this material.

 

The three types of information are presented: the introductory background material for the case; information regarding the tests that have been utilized in the assessment; and the testing results along with the standardization tables necessary to interpret them. I shall proceed in the order in which the information is presented in the assignment. The first thing we need to consider is the introduction to the assignment and what it reveals about this young man:

·        He is a 28-year-old white male.

·        He is in the last year of his B.Sc. in Biology – based on this we would estimate his number of years of formal education as 14.

·        Prior to the accident he was getting A’s and HD’s

·        Following the accident and his return to university he was achieving C and B grades.

 

Note that the assessments are conducted in the context of his university education, i.e. prior to his return to his studies and approximately a year later when he is considering graduate studies. As you would expect the impact of his injuries upon his ability to engage in tertiary education and the challenges he will face in doing so, will be uppermost in his mind.

 

Regarding the accident:

·        He was involved in a motor vehicle accident

·        He lost consciousness and remained so for 28 hours.

·        Period of retrograde amnesia is 10 minutes

·        Period of post-traumatic amnesia (PTA) is around three days

·        Glasgow Coma Scale of  7/15

 

The table below is adapted from the 1985 Central Nervous System Trauma Status Report edited by Donald P. Becker and John T. Povlishock. Prepared for the National Institute of Neurological and Communicative Disorders and Stroke, National Institute of Health, USA, 1985. It indicates categories for defining head injury severity. Based upon these criteria Mr. Gould’s traumatic brain injury would be classified as Severe (LOC=28 hours, GCS = 7, PTA = 72 hours).

 

 

 

 

 

 

 

 

 

 

He reports symptoms related to:

·        Headaches

·        Memory

·        Concentration

·        Word-finding

 

The data for only two tests have been provided. This would constitute a poor assessment if these were the only tests administered but the data presented is sufficient to enable you to test the hypotheses generated and get a taste for the psychological assessment interpretative process.

 

The Wechsler Adult Intelligence Scale – 3^rd Edition (WAIS-III) and the Wechsler Memory Scale – 3^rd edition (WMS-III) have been administered and some of the test scores from the initial and follow-up assessments have been provided. Data has also been provided based on the Wechsler Test of Adult Reading which can also be combined with demographic data to predict WAIS-II and WMS-III scores. Examination of this measure is not strictly necessary to complete the assignment (but the values it predicts certainly are), but this measure would be commonly employed in cases of this type to estimate premorbid or expected level of cognitive functioning.. The WTAR consists of progressively more difficult words which must be read aloud by the test-taker. Reading tests have been commonly employed to estimate IQ scores. As the words are roughly in order of reading grade level, those who have not been exposed to certain words will not know how to pronounce them. Thus performance on the test is a measure of the degree to which you have been exposed to infrequent or irregular words in the English language, something which is highly correlated with the number of years of formal education one has had. Some of the original studies with measures of this type  have indicated that, perhaps with the exception of severe impairment, declines in cognitive functioning are NOT accompanied by declines in performance on reading tests. Many clinicians, therefore use the measure as a way of examining word knowledge with the expectation that whatever problem you may have, has had little impact upon test performance. This is a round about way of saying that it is often used as an estimate of your abilities before you began suffering from your current difficulty. Individuals hypothesized to be currently experiencing a decline in cognitive functioning could be recognized as such by comparing where they should be based upon their WTAR + Demographic scores with where they are now! More about this later!

 

THE WAIS-III

 

            The WAIS-III is a cognitive/intellectual battery of tests designed to assess a variety of cognitive domains. Most notable in the battery is the omission of tests that evaluate memory beyond that of short-term recall. The battery consists of 14 subtests only 13 of which are customarily administered. The Object Assembly subtest has been included in the third edition of the battery to maintain consistency with its predecessor (the WAIS-R) and to permit the substitution of this test for a spoiled Block Design performance. In the case that you will be analyzing no data for Object Assembly has been provided.

 

            The nature of the subtests will be discussed shortly, but analysis of the WAIS-III focuses primarily around composite scores. Composite scores are combinations of subtests according to a particular theoretical framework. The two types of composites on the WAIS-III are IQ scores (which are based upon an historical framework) and Index scores (which are based upon an empirical framework). One of the things that can be somewhat confusing about the WAIS-III is that subtests are combined in ways that are similar but not identical for the two frameworks described. Let’s consider the IQ framework first.

 

            David Wechsler, the developer of the Wechsler batteries conceived of his test as a measure of overall cognitive/intellectual ability in the form of a score that included all of the subtests on the battery, the Full Scale Intelligent Quotient (FSIQ). This FSIQ could be divided into two sub-categories, the Verbal Intelligence Quotient (VIQ) which encompassed those measures that were administered verbally and required verbal responses,  and the Performance Intelligence Quotient (PIQ) which encompassed those measures that were administered visually and usually required a written, pointing, or object-manipulation response. So essentially, FSIQ represents an overall indication of performance and VIQ and PIQ are subdivisions that relate to the mode of input and output of information. This framework has been retained in the WAIS-III with IQ scores computed from the measures that have been traditionally used for this purpose.

 

            Factor analysis of the WAIS-III, however, reveals that rather than a dichotomous verbal and visual structure to the battery, four abilities or constructs are actually measured by the test. This factor structure forms the basis of the four Index scores. The Verbal Comprehension Index (VCI) consists of those subtests that best measure verbal comprehension and expression. The Perceptual Organization Index (POI) consists of those subtests that best measure an individual’s ability to process complex visual information and solve problems – in some ways a non-verbal reasoning measure. The Working Memory Index (WMI) consists of those subtests that best measure attentional abilities and the degree to which an individual can efficiently perform mental operations. The Processing Speed Index (PSI) consists of the subtests that best measure speeded visual information processing  - these measures all involve scanning of visual information and the rapid writing of responses on a page.

 

            As you are well aware, the more items that are included on a scale the higher the reliability is likely to be. Examination of the numbers of subtests that are included in each of the IQ and Index scores (in the table below) would appropriately suggest that higher internal consistency is found in composites that contain more subtests. Our approach to analyzing psychological test data is to proceed in a hierarchical fashion from the most reliable measures and work our way down to the least reliable measures. This means beginning with FSIQ, proceeding to VIQ and PIQ, then to VCI, POI, WMI, and PSI, and finally to the individual subtest level should this prove necessary.

 

The following is a general description of each of the WAIS-III subtests:

 

Vocabulary

This subtest presents 33 words orally to the test-taker and requires them to supply a dictionary style definition. It is the most reliable subtest in the scale and is the best measure of g (69% of its variance). It contributes to FSIQ, VIQ, and VCI composites and is generally considered a measure of word knowledge. This measure is heavily influenced by formal education and literacy.

 

Similarities

            This subtest consists of 19 word pairs. The test-taker’s task is to indicate how the two words are similar. More points are awarded for a more abstract relationship. It contributes to FSIQ,  VIQ, and VCI composites and is generally considered to be a measure of relational word knowledge. This measure is strongly influenced by education and literacy.

 

STRUCTURE OF WAIS-III COMPOSITE SCORES

 

No. of Subtests:                                                14

VIQ Subtests: (Number of Items)

                                                                        Information (28)

                                                                        Digit Span (15)

                                                                        Vocabulary (33)

                                    Arithmetic (20)

                                                                        Comprehension (18)

                                                                        Similarities (19)

PIQ Subtests:

                                                                        Picture Completion (25)

                                                                        Picture Arrangement (11)

                                                                        Block Design (14)

                                                                        Digit Symbol-Coding (133)

                                                                        Matrix Reasoning (26)

 

FSIQ = VIQ subtests + PIQ subtests

 

Factor Indices:

Verbal Comprehension

                                                                        Information (28)

                                                                        Vocabulary (33)

                                                                        Similarities (19)

Perceptual Organisation

                                                                        Picture Completion (25)

                                                                        Block Design (14)

                                                                        Matrix Reasoning (26)

Working Memory

                                                                        Digit Span (15)

                                    Arithmetic (20)

                                                                        Letter-Number Sequencing (7)

Processing Speed

                                                                        Digit Symbol-Coding (133)

                                                                        Symbol Search (60)

Arithmetic

            This subtest contains 20 items requiring progressively more demanding mental arithmetic. Factor analytic studies indicate that despite the arithmetic content that this is most commonly a measure of attention or working memory. This does not mean that people with dyscalculia or specific learning disability in arithmetic will not perform poorly on this measure. The content of the subtest requires mastery of relatively simple mathematical procedures such as percentages, averages, and probability. Consequently the influence of education on this subtest is greater than for the other measures that contribute to the Working Memory Index. This measure is included in the computation of FSIQ, VIQ, and WMI composites.

 

Digit Span

            This subtest also contributes to FSIQ, VIQ and WMI composites. The Digit Span task is the prototypical immediate memory or attentional task. On Digits Forwards, test-takers are required to repeat up to eight-digit sequences back in correct order as they were presented. The Digits Backwards task requires the digit sequences to be repeated back in reverse order. It is this second task that is, perhaps, more appropriately termed working memory as mental juggling of the number sequence is required to successfully complete the task. Characteristically, individuals recall on average 2 nire digits forwards than backwards. Less than 4% of the standardization sample recalled more digits backwards than forwards.

 

Information

            This 28 item subtest measures general knowledge through a broad range of questions about science, literature, geography, and historical events. This measure is highly correlated with educational achievement. This subtest contributes to FSIQ, VIQ, and VCI.

 

Comprehension

This 18 item subtest asks questions about social knowledge and awareness of socially appropriate behaviours and responses. A critical issue with this test is that it does not ask what you would do in a particular situation but rather what should you do. Similarly it asks not why you think something is so, but what we are taught are the reasons behind issues such as taxation, or the importance of a free-press. This subtest contributes to FSIQ, and VIQ. It does not contribute to any factor indices

 

Letter-Number Sequencing

This subtest contains seven items with three trials for each item. In some ways it is similar to Digits Forwards from the Digit Span subtest. In each trial the examinee is read a series of number and letters that have been placed in a random order. The examinee is required to reorder the numbers and letters and repeat them back in the correct ascending sequence with numbers first followed by letters. This subtest contributes only to the WMI composite. It does not contribute to any IQ scores.

 

 

Picture Completion

This subtest consists of 25 colour drawings of objects, people, and scenes where an important element is missing. The examinee is required to indicate what important element is missing from the picture. This subtest contributes to FSIQ, PIQ, and POI composites.

 

Digit Symbol-Coding

This subtest consists of a maximum of 133 items. The examinee is presented with a table containing the numbers 1 through 9 and symbols (simple line drawings) that are associated with each number. A template which contains 133 numbers (1 through 9) in a random sequence where the numbers are presented but the associated symbols have been omitted is presented to the examinee. Beginning with the first item, the examinee must fill-in as many symbols that go with each number in sequence in a two minute period. This subtest contributes to FSIQ, PIQ, and PSI composites.

 

Block Design

This subtest utilizes up to 9 blocks each with 2 red surfaces, 2 white surfaces, and 2 half red/half white surfaces. These blocks are employed by the examinee to replicate a design presented as a two-dimensional picture. Examinees manipulate the blocks and put them together in such a way so as to produce the same design with the top surfaces. The designs become progressively more complex and go from requiring 4 blocks to all 9 blocks in order to replicate the design.This subtest contributes to FSIQ, PIQ, and POI composites.

 

Picture Arrangement

This subtest consists of 11 series of line drawings that when placed in the correct order tell a story (allegedly humorous, but clients seldom laugh). The drawings are laid out in front of the examinee in an incorrect order and the examinees task is to reorder them in the correct arrangement. This subtest contributes to FSIQ, and PIQ composites. It does not contribute to any factor indices.

 

Symbol Search

This subtest was taken directly from the WISC-III and requires the examinee to look at two symbols and determine whether either symbol is present in a sequence of five symbols. There are 60 items each with two target and five test symbols. The examinee is required to verify the presence or absence of the symbols in as many items as they can in two minutes. This subtest contributes only to the PSI composite. It does not contribute to any IQ scores.

 

BACK TO THE CASE

 

                         When interpreting WAIS-III, WMS-III, or any psychological test data for that matter we proceed from the most reliable measures to the least reliable measures. This approach protects the clinician from being biased by an interesting, unusual, abnormal BUT unreliable finding. The most reliable measures are less prone to random variations and our decisions are more likely to be accurate if we base our interpretations on them. Now its time for me to get on to my soapbox. I will make a statement now that will seem radical but is eminently defensible through logic and basic psychometric principles. Measures of intelligence have no meaning in a clinical evaluation. This is not because the concept of intelligence does not exist (although I have seen little evidence to support it in my career) but rather that the measurement of intelligence necessarily presupposes that the individual being assessed is normal! Said another way, when you know that someone is normal then a test of intelligence may indicate that individual’s ranking relative to other normal individuals. This is not, however, the case when we use these tests in the clinical setting. Clinically, the Wecshler scales utilize tests of different cognitive abilities that are of interest to the clinician. In this context, scores such as Full Scale Intelligence Quotient FSIQ), Verbal IQ (VIQ), and Performance IQ (PIQ) cannot be construed as measuring intelligence in any way. To believe so would be to infer that a blind person is of low intelligence (VIQ = 100, PIQ = 47, FSIQ =74) (when it’s just that they can’t see any of the Performance subtests), a deaf person is of low intelligence (VIQ=48, PIQ= 100, FSIQ = 68) (when it’s just that they can’t hear the Verbal subtests), and a dead person is just profoundly intellectually impaired (VIQ = 48, PIQ = 47, FSIQ = 45). This last little gem is a consequence of the fact that despite the fact that the dead individual makes no responses to any of the questions receiving a raw score of 0 on every subtest, the scaled score associated with a raw score of 0 is 1 (surprisingly not 0). With 11 subtests constituting FSIQ, a dead person (or your favourite piece of lint if you prefer) receives a sum of scaled scores of 11 which corresponds to a FSIQ of 45.

 

                         So, having convinced you of the evils of the pragmatics of intelligence theory, what do FSIQ, VIQ, and PIQ mean? Good question. In the context of clinical assessment FSIQ represents overall performance on the majority of subtests. It functions essentially as a grand mean. It is of value to us because it has the highest reliability and has the potential for representing overall performance on the test. However, just as a grand mean may not actually be representative, we look to subgrouping of tests to determine whether the FSIQ is representative or not. VIQ and PIQ divide the subtests into two groupings based upon their input and output modalities. VIQ subtests are all administered verbally by the tester and answered verbally by the test-taker. PIQ subtests are all administered using visual stimuli and the test-taker can respond by pointing, writing, or manipulating objects. Two PIQ subtests may also involve giving verbal responses (Picture Completion and Picture Arrangement) but the scoring of each of these tests does not rely on a verbal response. These subgroupings are historical rather than empirical. By this I mean that the existence of VIQ and PIQ as scores was based upon David Wechsler’s belief that these divisions would be meaningful and not based upon research findings that support this distinction. Empirical studies (factor analyses) of the WAIS-III in fact support four composites or indices and these form the true basis of interpreting the WAIS-III in the clinical setting. What is lucky for us is that the four constructs, in principle, reflect a subgrouping of each of VIQ and PIQ into two smaller divisions. I say, in principle, because in each case one of the subtests omitted from the IQ score is included in an index score (Letter Number Sequencing is omitted from VIQ but is part of WMI, and Symbol Search is omitted from PIQ but is part of PSI).

 

                         So in proceeding from the most reliable to the least reliable measures we begin with FSIQ, proceed to VIQ and PIQ, and then to the Index scores VCI, POI, WMI, and PSI. After, and only after, that you can proceed to consider individual subtests if they address relevant hypotheses or in order to examine the integrity of index scores.

 

            Enough chatting, let’s begin looking at the table below. The primary purpose here is to teach you how to read the information in the tables. Consistent with the approach described above let’s start with FSIQ. Mr. Gould obtained a FSIQ Standard Score (SS) of 108. The next column, headed %ile, indicates the relative ranking of this score in the standardization sample. Standard scores have a mean of 100 and a standard deviation of 15, so this score is just over half a standard deviations above the mean which corresponds to the 70^th percentile. How would we say this in language understandable to more people than just myself and any other unfortunate soul who lives and breathes psychometrics? Standard scores are all age-adjusted so we could describe this as “Mr.Gould’s overall level of cognitive functioning is estimated to fall within the Average range and is as good as or better than 70 percent of people of his age.” Where did I get the “Average” statement from? The table titled “WAIS-III QUALITATIVE DESCRIPTIONS FOR IQ SCORE RANGES” in “TABLES TO ASSIST IN ANALYSING TEST DATA FOR THIS CASE” which accompanies the assignment.

 

The next two columns contain information regarding the 90% confidence interval at the time of testing. There are two numbers the low end of the range (Lo) and the high end of the range (Hi). These numbers tell us the range in which we can be 90 percent sure that this person’s true ability lies. Remember that the score you compute on a psychological test is just an observation. We need to determine the range in which the underlying ability we are trying to measure is likely to truly fall. This is never one score but a range of scores. So while Mr. Gould obtained a FSIQ of 108, these two columns tell me that I can be 90% sure that the score of 108 reflects an ability somewhere between 104 and 111.

 

                The last two columns contain the 90% confidence interval for a score on that measure at retest. The Lo and Hi columns indicate the lower and upper limits of the range. So we can be 90% sure that because Mr. Gould got a FSIQ of 108 at his initial testing that when we retest him it should fall somewhere between 103 and 113. The implication of this is that if we tested Mr. Gould some time later and he got a FSIQ of less than 103 or more than 113 we would have direct evidence that his score has changed (declined if less than 103, or improved if greater than 113). As you can see these last two columns would only be used if a second testing had been performed – more on this later!

 

 

So, with the information above we can describe Mr. Gould’s performances on the three IQ measures and four index scores of the WAIS-III indicated in the table above. Remember that only other psychologists are going to know all this jargon, so rather than referring to VCI you would talk about a measure of his verbal comprehensive and expressive abilities. Rather than WMI you would say something like a measure of attention, concentration, and the ability to efficiently perform mental operations.

 

Now the next topic relates to the detection of abnormality in the profile of scores and uses the two tables below. We have already discussed the idea of FSIQ reflecting Mr. Gould’s overall level of functioning. To test this idea we need to consider that FSIQ may not be made up of 11 homogeneous test scores but rather systematic differences that average out to the FSIQ score. The first way of addressing this is to examine the subsets of FSIQ, VIQ and PIQ. If VIQ and PIQ differ significantly, then FSIQ does not represent a homogeneous level of functioning. VIQ for Mr. Gould is 111, and PIQ is 104. The difference between these two measures is (111-104) 7. If we consult the table of significant differences between WAIS-III composite scores for Mr. Gould’s age group we find that an 8.27 difference would be significant at p=.05. The 7 point difference is less than this so we can assert that VIQ and PIQ do not differ significantly.

 

Differences Between IQ Scores and Between Index Scores Required for Statistical Significance at the .05 Level for the 25 to 29 Age Group

 

But what does this mean? A test of significance is testing the likelihood that the two scores came from the same distribution. Do not be swayed by the apparent size of the difference. Just looking at the significance table for the different WAIS-III composites shows differing values from 8.27 to 13.14. These values are different because of the differences in reliability of the measures involved. The more reliable two measures are, and the more inter-correlated they are, then the less a difference needed for significance. Just remember that good reliability results in things being more easily detected and poor reliability results in things being much harder to detect. Back to our original question here, what does it mean that VIQ and PIQ are significantly different? The easiest way to understand this is to turn it around. What is the null hypothesis? That VIQ and PIQ are identical (no difference between them). A significant difference means we reject the null and infer that the two numbers are not the same. Nothing more, nothing less. Knowing that VIQ is not significantly different from PIQ and means that we can infer that Mr. Gould has similar verbal and visual/graphomotor abilities (remember VIQ and PIQ reflect the modalities of the tests).

 

Disappointed? Hoping for more? What you really probably wanted to know was whether or not the difference is clinically meaningful! This is not addressed by significance (although this is not always true) but rather by abnormality. We address abnormality by doing a “head-count”. This is the testing equivalent of asking “OK! Hands up all those people who did …”. How common is a 7 point difference between VIQ and PIQ in a person who has a FSIQ in the Average range.  That question can be answered by the table below – 27.3% of the standardization sample.

 

One-Tailed Frequencies of Differences Between WAIS-III IQ and Index Scores for Individuals with FSIQ of 90 to 109

 

This raises the next big question – how rare is rare? All clinicians who use psychological tests have to ultimately make a decision about this. I can tell you what I do and why. Other decisions are not wrong but like anything else in life there are consequences to what we decide. For tests of significance I use p<.05 as the standard for a statistically significant difference with a two-tailed test. For abnormality, I consider that anything that occurs with a frequency of 1 in 20 or less (5%) is sufficiently rare to call the behaviour abnormal. I will confess, however that analyses that have a frequency of 6-10% are of particular interest to me. I am not permitted to change my criterion when I have a behaviour that occurs with a frequency between 6 to 10%, but what I can do is keep it in the back of my mind and pay particular attention to any opportunities that may arise to test the hypothesis (i.e. is it abnormal or normal). One other comment here, abnormality refers only to infrequency or rarity of the behaviour, it does not tell you whether or not the behaviour is impaired. For example, an accountant is likely to be highly skilled at mental arithmetic so his score on the WAIS-III Arithmetic subtest is likely to be abnormally higher than other scores. This abnormality will be detected during data analysis, but does not signal impairment. Some abnormalities are good – we call them skills, some abnormalities are bad- we call them deficits. You will need to determine when an abnormality is signaling a skill versus an impairment. The final criterion is the confidence interval applied to test scores. I use 90% confidence because I am happy to have an error rate of 5% at either end of the distribution. Too lax a criterion (68%) will result in too narrow a band of scores while too strict a criterion (99%) will result in a band that is too wide to be of use (i.e. I can be 100% sure that the FSIQ score you got on the test was somewhere between 45 and 155! Well, duh!).

 

            A second point. Some people (not me, but you know, “other people”) might ask: If we care about abnormality so much, why don’t we go straight to those tables and leave out the test of significance? Ahhhh, the impetuosity of youth! Remember, the score we obtain on the test is only a sample of behaviour from the individual. It is no more necessarily characteristic of that individual than your blood pressure at midday is representative of your blood pressure overall. There will certainly be a relationship, but this is why we have confidence intervals – to tell us the likely range in which a person’s score is likely to fall when we have only one observation. So when we ask if a VIQ of 111 and a PIQ of 104 are significantly different we are implicitly taking into account the less than perfect reliability of our measurement. Although this does not make it easier, it is understandable if we stop talking about scores as if they are real things. The question we are really asking is: How likely is it that a VIQ which I am 90% sure falls somewhere between 107 and 115 is the same level of performance as a PIQ which I am 90% sure falls somewhere between 98 and 109? Can you see that the assumption that the real difference between VIQ of 111 and PIQ of 104 is 7 must be tested. Since this difference is less than that necessary to meet our criterion of significance we accept the null and assert that for all intents and purposes the difference is 0 (i.e. we cannot reject the null hypothesis). So what does this have to do with the point? When we look up how frequent a difference is (like 7) it is based upon the assumption that 7 is the right number to reflect the difference! The significance test, in this case for VIQ and PIQ, tells us that 7 is not the best difference and it could well be 0. Consequently when the test of significance indicates a non-significant finding you do not know what difference value to look up – it certainly is not 7 and probably should be 0! When we do find a significant difference as with VCI and WMI in the case of Mr. Gould then the 21 point difference (which exceeds the critical difference of 9.67 necessary for significance) is assumed to be the best estimate of the difference between these two measures, and we happily look up the frequency of a difference of 21 between VCI and WMI and find that it is rare - 4.8%! What this all leads to a simple rule: If the difference is significant then look up the frequency of the difference! If the difference is not significant DO NOT look up the frequency of the difference – it is not abnormal! You will note that despite this rule I have still supplied you with the frequencies for all differences – this is to prevent you from assuming that the only information I provide is meaningful!

 

            We now repeat this process for the Index scores of the WAIS-III. Six comparisons can be made: VCI with POI, VCI with WMI, VCI with PSI, POI with WMI, POI with PSI, and WMI with PSI. Essentially we are looking for significant differences to indicate where performance levels differ and then for those, AND ONLY THOSE, statistically significant differences we then examine the frequency of the difference in order to detect abnormal differences. Note that many of the comparisons are statistically significant, indicating that this individual has distinctly different performance levels for each of the cognitive domains assessed on the WAIS-III: verbal comprehension, visual organization, attention/concentration, and speed of information processing. Notice also that many of the comparisons with WMI ands PSI are associated with frequencies of less than 5% in the standardization sample. This is revealing a consistent pattern – Mr. Gould’s attentional abilities (WMI) and processing speed (PSI) are abnormally lower than his verbal (VCI) and visual organisational abilities (POI).

 

            The next table, below, can be used to examine the individual subtests for relative strengths and weaknesses. The reasoning behind this analysis goes as follows. A relative strength is a performance on a subtest that is comparatively higher than other subtests. Similarly a relative weakness can be seen on those measures that are comparatively lower relative to other measures. Each subtest is usually compared to the mean of subtests. The question is how do we compute the mean? This goes back to the discussion regarding the representativeness of FSIQ. Simply put, if there is no difference between VIQ and PIQ (as in this case) then the average of all the subtests administered can be used. If there is a difference between VIQ and PIQ then separate means must be computed for verbal and performance subtests.

 

Differences Between Single Subtest Scaled Scores and Mean Scaled Score at the .05 Level of Statistical Significance and Magnitude of Difference Found in 5% of the Standardisation Sample

	Verbal	Subtests	Performance	Subtests	All	Subtests
Subtest	p<.05	5%	p<.05	5%	p<.05	5%
VO	2.10	3.00			2.30	3.38
SI	2.77	3.29			3.12	3.69
AR	2.63	3.57			2.95	3.85
DS	2.40	4.43			2.67	4.62
IN	2.34	3.29			2.59	3.69
CO	2.96	3.57			3.35	3.58
LNS	3.16	4.29			3.60	4.38
PC			3.16	3.86	3.46	4.31
CD			3.04	4.29	3.31	4.46
BD			2.94	3.71	3.19	3.92
MR			2.60	3.71	2.75	3.85
PA			3.75	4.14	4.19	4.46
SS			3.54	3.86	3.93	4.23

 

The mean for the thirteen subtests is 10.6. We now compare each subtest with this overall mean. Subtracting the mean from each of the subtests yields a pattern of positive and negative difference scores. A positive value means that the subtest is above that individual’s mean score, while a negative value indicates a lower performance. Comparison of these differences with those in the table above indicate which are significantly different from their respective means. Note that the two rightmost columns are used in this case because VIQ and PIQ do not differ significantly. The first thing you are looking for is whether or not each number differs significantly from it’s mean, indicated by an absolute difference greater than or equal to the cut-scores indicated in the table (we are only using column 6 here to detect significant differences). For those measures that are significantly different from their respective means, those that are positive are “relative strengths (S)” and those that are negative are “relative weaknesses (W)”. These findings can be used to describe where Mr. Gould’s strengths and weaknesses lie in terms of the behaviours assessed by each subtest. You can also examine column 7 to see how unusual or infrequent a significant difference is. For example the Information subtest has a scaled score of 16 which is 5.4 points above the mean of subtests. This difference is significant as 5.4 is greater than the critical value indicated for p<.05 in the table (2.59). This indicates that for Mr. Gould the Information subtest is a relative strength. Looking at column 7 we can see that 5% of the standardization sample had differences between Information and the mean of 3.69 or greater. From this we can infer that Mr. Gould’s difference of 5.4 is highly unusual and would not be expected to occur as a result of normal random variation.

 

Verbal Subtests	SS	SS-Mn	S/W	Performance Subtests	SS	SS-Mn	S/W
Vocabulary	13	2.4	S	Picture Completion	13	2.4
Similarities	10	-0.6		Digit Symbol-Coding	7	-3.6	W
Arithmetic	11	0.4		Block Design	12	1.4
Digit Span	9	-1.6		Matrix Reasoning	10	-0.6
Information	16	5.4	S	Picture Arrangement	11	0.4
Comprehension	12	1.4		Symbol Search	6	-4.6	W
Lett.-Num. Seq.	8	-2.6
				Mean for All Subtests	10.6

 

            A note about the subtest scores. The table below reproduces the subtests scores provided for Mr. Gould. Each subtest is named and two numbers are provided. In the first column of numbers the acronym SS in this case stands for Scaled Score. Scaled scores have a mean of 10 and a standard deviation of 3 and are adjusted for the age of the information (usually termed age scaled scores). The second column of numbers indicates the percentile rank of the scaled score. For example, Mr. Gould’s score on the Vocabulary subtest was 13 which indicates that his knowledge of the meaning of words is as good as or better than 84% of people of his age.

 

                        

            The next table asks about the status of Mr. Gould’s WAIS-III scores in a different way. While the previous analyses have focused on the level of performance amongst his cognitive abilities, the next analyses asks the question “Based on what we know about you, where should your scores lie?” Based on certain types of information, the clinician attempts to construct a pattern of hypothetical scores which represent what the person would look like if they had no disorder or condition, or in other words what the person would have looked like before the accident.; You should immediately recognize that if Mr. Gould had been tested on the WAIS-III and WMS-III before his accident we would not need to do this at all. We would compare directly his current scores with the earlier testing. Unfortunately for psychologists, people do not routinely engage in psychological testing just in case they may have an accident (although the idea certainly has some merit!). Such circumstances are viewed as “pure gold” or “mana from heaven” when they occur, but in the vast majority of cases we are forced to estimate what a person’s scores would likely have been before the accident. This is called estimating premorbid levels of functioning.

 

            I have tested the client and need a value from his past that will accurately tell me what he was like before his difficulties. There are a number of ways of getting such values. The first is to use information that is not related to the pathology such as demographic characteristics of the individual. Such an approach asks the question “What should your FSIQ be given that you are a male in your 20’s engaging in university level education. This approach estimates premorbid functioning using demographic variables alone.

                         A second approach, the one used in this case, combines demographics with a test that we believe is unlikely to be affected by whatever is wrong with Mr. Gould. Although this is a current measure we are using it to “guesstimate” what he would have looked like before his difficulties. For this approach we use the Wechsler Test of Adult Reading (WTAR) along with demographic variables to estimate Mr. Gould’s WAIS-III performance (and later his WMS-III performance). So, Mr. Gould’s score on the WTAR along with his gender, age, and educational level estimate that his VIQ should be 113. His obtained score of 111 does not differ significantly from this estimate (for a difference to be significant it would have to be 8.1 or larger) indicating that his current VIQ falls where we would expect it to. Remember you don’t need to look up frequencies for non-significant differences. Accordingly only WMI and PSI are of interest. The 16 point difference between Mr. Gould’s actual WMI score of  95 and his predicted score of 111 is significant and somewhat rare, with differences of this magnitude found in only 5% to 9% of the standardization sample. The 23 point difference between his actual PSI of 81 and his expected value of 104 is significant and extremely unusual, occurring in only 1% of the standardization sample. This analysis further supports the contention that Mr. Gould’s WMI and PSI scores are abnormally low and are likely to reflect cognitive impairment.

 

 

Okay, before we go on to the memory testing let’s take stock.  Ask yourself the following questions (and answer them!):

·        What range of cognitive/intellectual functioning best represents Mr. Gould’s current abilities on the WAIS-III?

·        Does this level suggest impairment?

·        Were there any abnormalities detected in the WAIS-III scores?

·        What cognitive domains were these abnormalities related to?

·        Do these abnormalities reflect skills or impairments?

·        Is there evidence at this point for deterioration in Mr. Gould’s cognitive/intellectual functioning from an estimated premorbid state?

 

And now? On to the WMS-III!

 

STRUCTURE OF THE WMS-III

 

            The third edition of the Wechsler Memory Scale contains a number of subtests designed to assess both short-term and long-term memory functioning. The nomenclature of measures on this test can be confusing and mixes term from classical and modern memory theory. Unlike the WAIS-III there is no one overall measure comparable to FSIQ although there are measures similar to VIQ and PIQ. In this description of the test I will only focus on those measures necessary to generate the respective Index scores and will not include the optional subtests.

 

            The WMS-III essentially measures three abilities: the ability to recall information shortly after its presentation (Immediate Memory), the ability to recall this same information after a 20 to 30 minute delay (General Memory), and the ability to attend and concentrate (Working Memory). Within each of the first two measures (Immediate and General Memory) there are subdivisions that are based upon whether or not the test is verbally or visually administered and whether or not the examinee had to recall or only recognize the information that was presented. In the case of this assignment these subdivisions are not relevant to either the analysis or interpretation. They will be discussed here for the sake of thoroughness.

 

The Immediate Memory Index consists of two verbal subtests (comprising Auditory Immediate) and two visual subtests (comprising Visual Immediate). The first verbal subtest, Logical Memory, involves the presentation of two stories which are then to be repeated back by the examinee in as much detail as possible. The second story is presented twice and recall is tested to get a gross measure of learning. The second verbal subtest, Verbal Paired Associates, presents 8 pairs of words which are read aloud to the examinee. Each word pair is an uncommon pairing of words (such as flower-paperclip) and recall is tested by the examiner supplying the first word of the pair (flower) and the examinee must supply the second word of the pair (paperclip). Four trials are administered with the examiner rereading the list of word pairs each time and testing cued recall.

 

            The two visual subtests are Faces and Family Pictures. In Faces, the examinee views 24 photographs of faces and is then asked to determine which 24 out of a further 48 faces they have seen before. The Family Pictures subtest introduces the examinee to seven characters (Grandmother, grandfather, father, mother, son, daughter, and dog – I know, I know, blatant discrimination against cat people!). The examinee is then shown four scenes in which different family members are doing different things in different parts of the picture (Maybe this is why they didn’t use a cat – the cat would be doing the same thing in  all four pictures – nothing!). After seeing all four scenes, the examinee is required to indicate who was in each picture, their location, and what they were doing.

 

            The General Memory Index consists of the delayed recall trials (administered approximately 20 to 30 minutes after the immediate recall) of these same subtests. Auditory Delayed consists of delayed recall of the Logical Memory stories and the Verbal Paired Associates. Visual Delayed consists of the delayed trials of the Faces and Family Pictures tests. A third measure Auditory Recognition Delayed consists of the recognition trials of Logical Memory and Verbal Paired Associates that are presented after their delayed recall. General Memory Index is therefore made up of Auditory Delayed, Visual Delayed, and Auditory Recognition Delayed measures.

 

            The Working Memory Index consists of the Letter-Number Sequencing subtest of the WAIS-III and a visual span task called Spatial Span. This task is very much like digit span except that rather than repeating forwards or backwards number sequences read by the examiner, the examinee taps out (forwards or backwards) a series of patterns tapped out by the examiner on a board with ten blocks in various positions. One thing to note is that the Letter-Number Sequencing score used in this Working Memory Index is not from a second administration of the test but rather is exactly the same number as was used in the WAIS-III. I know it seems insane, the reasons are too complicated to discuss here – sufficient to say that the WMI on the WAIS-III and the WMI on the WMS-III are far from independent assessments of the same construct.

 

 

BACK TO THE CASE

 

Let’s now look at Mr. Gould’s scores on memory testing. I am going to make this section a little easier for you by reducing the amount of information that you have to process. Remembering our interpretative principle of examining the most reliable information first, our analysis will be focused on the Immediate Memory, General Memory, and Working Memory indices. All of the other indices listed in the assignment are subsets of the first two indices that relate to visually and verbally administered tests. As none of the analyses reveal anything of interest regarding the verbal versus visual presentation I will remove them from the data presented below. All of the analyses that we will be conducting here can be applied to the measures I am removing and you are more than welcome to perform them – I am only removing them because they have no impact on this particular case (which I know because I analysed the data, heh heh!).

 

            Mr. Gould’s scores on the WMS-III were 89 for the immediate recall of information, 77 for the recall of information after a 20 to 30 minute time delay, and 91 for his ability to attend and concentrate. These performances were as good as or better than 23%, 6%, and 27% of people of his age and corresponded to performances in the low average, below average, and average ranges respectively.

 

 

There are three comparisons that can be made here, the most important of which is the IM comparison with GM. This addresses the issue of whether or not Mr. Gould’s memory performances are detrimentally affected by increasing the delay between presentation and recall. There is a 12 point difference between Mr. Gould immediate (IM) and delayed (GM) recall. This difference is significant (critical value of 11.8) and abnormal (estimated to occur in 4.4% of the standardization sample). This indicates that Mr. Gould’s ability to recall information after a time delay is abnormally poor relative to his ability to recall information when it is first presented. The comparisons of IM and GM with Working Memory (WM) also indicate a significant difference but only for GM and WM (a 14 point difference). This difference while significant is not considered abnormal as such discrepancies are found in almost 19% of the standardization sample.

 

Differences Between WMS-III Primary Index Scores for Statistical Significance at the .05 Level of Significance for the 25 to 29 Age Group

 

Frequencies of Differences Between WMS-III Primary Index Scores

 

While these findings strongly indicate impairment in delayed recall (GM) there is another method that we can use to determine whether or not his immediate memory is where it should be. We can address this issue in two ways: first, by asking whether or not Mr. Gould’s WMS-III scores are normal for a man with his intellectual abilities (as indicated by his FSIQ), and second, by asking what we would expect a man of his age, education, and reading ability to obtain on the WMS-III. The tables below (again altered to remove the unnecessary comparisons) address these questions.

 

Based upon Mr. Gould’s FSIQ of 108 we would expect that his predicted IM, GM, and WM scores would all be 105. He actually obtained 89, 77, and 91 for these measures yielding differences of 16, 28, and 14 between estimated and obtained scores respectively. Only the GM and WM differences are significant and indicate an abnormally low GM (frequency of 1-2%) and a suspiciously low WM (frequency of 10%). This supports that Mr. Gould’s scores on delayed memory are abnormally low for a man of his cognitive/intellectual abilities.

 

Comparisons of WAIS-III and WMS-III Composites Using Predicted Difference Method (Based Upon a FSIQ of 108)

 

 

The second approach uses the WTAR+demographics to predict expected WMS-III scores. Based upon his WTAR, age, gender, and education we expect him to have an IM score of 102, GM score of 104, and WM score of 108. Comparison with his actual scores of 89, 77, and 91 yield differences of –13, -27, and –17 respectively. While all three scores are significantly lower than would be expected, only GM is clearly abnormal (by my criteria with a frequency of 2-4%), while WM is suspiciously low (frequency of 5-9%).

 

Premorbid Estimates of WMS-III Composites Using WTAR + Demographic Data

Let’s consider those questions again with respect to the WMS-III:

·        What range of memory functioning best represents Mr. Gould current abilities on the WMS-III?

·        Does this level suggest impairment?

·        Were there any abnormalities detected in the WMS-III scores?

·        What abilities were these abnormalities related to?

·        Do these abnormalities reflect skills or impairments?

·        Is there evidence at this point for deterioration in Mr. Gould’s memory functioning relative to his other cognitive/intellectual functioning indicated on the WAIS-III and/or compared to his expected level of performance?

 

Hang in there! Almost done!

 

We can now look at the testing from eleven months later. This analysis is the most straightforward but it requires you to again work with information from multiple tables. Again for simplicity’s sake I will remove the irrelevant measures.

 

            Now this is where those retest confidence intervals from the first testing come into the analysis. The retest confidence interval allows us to determine whether or not scores on the second testing have deteriorated, stayed the same, or improved. You will remember that this determination is critical to our differentiating among the alternative hypotheses.

 

Results of Psychological Testing Eleven Months Later

 

We will now generate a table that combines the information from our first testing and second testing. We need the 90% RETEST confidence bands from the FIRST testing, the actual scores from the  SECOND testing and then consideration of where these scores fall relative to the retest bands. Scores that fall within the retest band are unchanged. Those below it, have declined, those above it have improved.

 

This analysis indicates that Verbal Comprehension, Perceptual Organisation, and General Memory have improved to a degree where it cannot be attributed to random variation in test scores.

 

 

 

To better understand the implications of these changes, let’s take a look at the retest  WAIS-III and WMS-III data.

 

 

VIQ for Mr. Gould on retest is 115, and PIQ is 110. The difference between these two measures is (115-110) 5. The table of significant differences below indicates that an 8.27 difference would be significant at p=.05. As the 5 point difference is less than this, we can assert that VIQ and PIQ do not differ significantly.

 

Differences Between IQ Scores and Between Index Scores Required for Statistical Significance at the .05 Level for the 25 to 29 Age Group

 

What about the Index Scores?

 

Clearly, Mr. Gould’s attentional abilities (WMI) and processing speed (PSI) are abnormally low relative to his verbal abilities. This ties in with the retest comparisons in that his VCI improved from test to retest but his WMI and PSI did not. If you think about it, this suggests that although we did not detect it in the test data, there must have been some lower than normal scores in VCI that have now recovered on retest (take a look at Similarities and remember Mr. Gould’s concern at not being able to do this test as well as he expected). By the way, there are analyses which would have picked this up in the test data, but I thoughjt you already have enough to deal with, without adding another three or four analyses to the assignment. No don’t thank me,. it’s all part of the service! Similarly, POI-PSI is abnormally low (2.5%) indicating that Mr. Gould’s processing speed (PSI) is still abnormally low relative to his other visual abilities (POI). As with VCI, POI also improved between test and retest and for similar reasons (take a look at Matrix Reasoning!)

 

So, we can tell a number of things at this point. Initially Mr. Gould demonstrated problems with his attentional abilities and processing speed. His verbal and perceptual organizational abilities appeared to fall within the range that would be expected for this man. Retest data indicate that his attentional abilities and processing speed have not substantially improved although his verbal and visual processing scores have (primarily due to improvement in the two abstract reasoning/problem solving tasks).

 

What about memory? The tables below indicate that there are now no significant differences between Mr. Gould’s immediate recall (IM), delayed recall (GM), and attentional abilities (WM). This suggests that Mr. Gould’s ability to retain information over a thirty minute period is commensurate with his ability to retain  it shortly after presentation. This is also consistent with the test-retest finding that Mr. Gould’s delayed recall has improved.

 

 

 

Differences Between WMS-III Primary Index Scores for Statistical Significance at the .05 Level of Significance for the 25 to 29 Age Group

 

Frequencies of Differences Between WMS-III Primary Index Scores

To understand how his memory now is relative to his intellectual abilities we need to compare IQ with memory scores. Note that because Mr. Gould’s FSIQ has gone up from 108 to 113 we must now use a different normative table:

 

Comparisons of WAIS-III and WMS-III Composites Using Predicted Difference Method (Based Upon a FSIQ of 113).

 

 

Notice how, despite the improvement in GM from test to retest, compared to his intellectual abilities it is still abnormally low. Thus, while his delayed recall (GM) has improved over the eleven months, it is still abnormally low compared to his other intellectual abilities.

 

All that is left now is to write an interpretation of this data. Unfortunately, since this is what the assignment is about I cannot help you further. In thinking about the data and Mr. Gould’s plight, consider what sort of question each analysis is capable of addressing:

·        The percentiles of each of the subtests and indices permit you to describe Mr. Gould’s current level of functioning.

·        The WAIS-III subtest analysis permits you to detect abilities or performances which are particularly strong or weak for Mr. Gould.

·        The analysis of WAIS-III indices permits you to determine whether or not his cognitive abilities are related in the way that they are in most people, or whether there is something unusual about his pattern of abilities.

·        The analysis of WMS-III indices asks the same of the memory indices – “Are they put together in the way that they are for most people?”

·        The analysis comparing WAIS-III with WMS-III permits you to determine whether or not Mr. Gould’s memory abilities are consistent with his other intellectual abilities.

·        The analyses involving the WTAR + demographics permits you to answer the question “Where should Mr. Gould’s scores be?”  Comparison of these values with those he actually obtained provides insight as to which abilities may have been detrimentally impacted upon by his injuries.

·        The Retest Confidence Intervals permit you to examine which measures have changed over the eleven month interval between tests.

·        The intregration of all of these analyses will assist in constructing a picture of not only what Mr. Gould’s cognitive abilities were like three months after his motor vehicle accident and again eleven months later, but also how these abilities may have changed over time. Recognition of these strengths and weaknesses will be critical in advising Mr. Gould on his educational and career aspirations.

 

This is a demanding assignment because it asks you to grasp a lot of information to which you have not been formerly exposed. However, this is a good model of how 21^st century assessors go about analysing and interpreting psychological test data. Before you feel too hard done by, consider the small number of variables that you have to consider in this case from only two tests. In a routine assessment in my own clinical practice more than twenty tests are customarily administered each with many scales and subscales. The process you have learned here is essentially the same as that applied to analyse twenty tests or two hundred variables, it just becomes progressively more complicated with the addition of more measures. The other warning I should give is that this case has been structured and simplified to illustrate certain principles. Most psychological assessments are highly complex and full closure is seldom achieved. The process is the same as in this case, but the outcome is rarely as neat.

 

 

 

 

This concludes the ancillary materials provided to assist you in completing assignment 2. Enjoy!

 

Dr. Graeme Senior

Senior Lecturer

Department of Psychology

University of Southern Queensland