The Category Test History development and future directions PAR by alicejenny


Nick A. DeFilippis & Jonathan N. Dodd
Georgia School of Professional Psychology
Early Tests of Concept
 Ach (1921), a German researcher developed a
  means of studying concept formation
  (Sakharov, 1930; Hanfmann & Kasanin, 1937).
 Vygotsky’s variation (Hanfmann & Kasanin,
 Hanfamann & Kasanin (1937) translated the
  Vygotsky test into English and identified 3
  basic factors of concept formation.
Early Tests of Concept
 Weigl Color-Form Test (1941) consisted of 12
  cardboard figures that could be grouped
  together according to their colors or forms.
 The Goldstein and Scheerer Sorting Test
  (1941) required the subject to sort 7 of 33
  common objects together and describe how
  they are alike and deliniated abstract concept
  formation into 3 levels of graded
  sophistication: Concretistic, Functional, and
Early Tests of Concept
 Ward Halstead developed an object sorting
  task similar to Goldstein and Scheerer’s
  known as the Halstead Object Sorting Test
 Straus & Werner (1942) conducted the first
  studies with this instrument comparing
  children with acquired brain injuries with
  those with mental retardation.
Birth of the Category Test

 Due to variable success with the Halstead
  Object Sorting Test, Ward Halstead
  developed the Halstead Category Test,
  designed to measure “grouping behavior.”
 The initial version of the Category Test
  contained 9 subtests totaling 360 items.
 The test was administered via an apparatus
  that contained a rotating drum and reading
Evolution of the Category Test

 During the 1940’s and 1950’s the method for
  administering the CT changed.
 Incorporated the use of slides, the number of
  items was reduced to 208, and subtest
  reduced to 7, and it was placed in a console.
 Variations of test administration included
  shortened versions, paper-and-pencil forms,
  a card version, a computer version, a booklet
  version, and an adapted version based on
  item response theory.
Evolution of the Category Test

 The Booklet version of the Category Test,
  known as the “Booklet Category Test” (BCT)
  has enjoyed much popularity due to it’s ease
  of transportation and administration, as
  compared to the original version using a
  cumbersome and heavy console.
 The booklet version consisted of 208 8 x 1
  paper cards containing stimuli identical to the
  original and bound into two separate
Shortened versions of the
Category Test
 Charter, Swift, and Blusewicz (1997) developed a
  shortened version of the CT composed of all
  items in subtests 1 through IV and 20 items from
  Subtests V from the original CT.
 Charter et al.’s sample consisted of 67 diffusely
  brain damaged and 109 non-brain damaged
  hospital patients matched for age and
 Limitations include limited demographic data for
  the control group and risk of selection bias.
Paper-and-pencil version

 Adams and Trenton (1981) modified the
  original CT into a paper-and-pencil version in
  efforts to reduce the cumbersomeness.
 They also created an alternative version of
  their paper-and-pencil version with good
 Adams and Trenton noted significant practice
  effects; thus, repeat testings between the
  two versions are not recommended.
Booklet Category Test

 Research has shown that the BCT correlates
  significantly with the original Halstead Category Test
  (McCampbell and DeFilippis, 1979).
 Comparing Category Test performance of 30 chronic
  alcoholics to that of 30 normal control college
  students, with order of test administration (original
  Halstead slide vs. booklet version) varied in
  administration, strong correlations of .913 and .804
  were found between normal and alcoholic
  participants respectively.
 Significant practice effects b/w first and second
  administrations of the CT in both normal and
  alcoholic groups.
Psychometric Properties of the
 The CT’s overall internal consistency is
  approximately 97% (Lopez, Charter, &
  Newman, 2000).
 Split-half reliability is 0.98 (Shaw, 1966).
 The CT shows significant age and education
  effects (Heaton, Grant, & Matthews, 1991).
Psychometric Properties

 Most researchers agree that the CT is a
  multidimentional instrument measuring
  several independent but complimentary
  cognitive abilities including:
   Concept formation, abstract reasoning, nonverbal
    problem solving, and attention (Johnstone,
    Holland, & Hewett, 1997)
   Learning (Laatsch & Choca, 1994)
   Judgment (Lopez, Charter, & Newman, 2000)
Properties:Construct Validity
 Research indicates that the BCT correlates
  with measures of intelligence, memory, and
  nonverbal problems solving (Allen, Goldstein,
  & Mariano, 1999).
 Lansdell & Donnelly (1977) conducted a factor
  analysis that indicated that the CT loaded
  highly on visuomotor factors, as well as WAIS
  Performance subtests, Block Design, Picture
  Arrangement, and Object assembly.
Psychometric Properties:
Construct Validity
 Leonberger, Nicks, Goldfader, & Munz (1991)
  found that the Category Test loaded on tests
  of visual, but not verbal memory.
 Jonstone, Holland, & Hewett (1997) also
  conducted a factor analysis finding that
Psychometric Properties:
Construct Validity
 Johnstone, et al. (1997) fourn that there was
  not a single CT factor of abstract reasoining.
  Rather, they identified three different factors
  (cognitive demands) including:
   Symbol Recognition/Counting
   Spatial Positional Reasoning
   Proportional Reasoning
Psychometric Properties:
Lateralization Effects
 Dodd, McDermott, Goldstein, & DeFilippis
  (2008) found that the Spatial Positional
  Reasoning (SPR) Index correlated with
  patients with right hemisphere lesions and
  Performance IQ scores.
 In contrast, the Proportional Reasoning (PI)
  index correlated with patients with left
  hemisphere lesions and their Verbal IQ
Psychometric Properties: Effort

 Using item analysis, Laatsch & Choca (1991)
  found that all items from Subtests I and II did
  not discriminate between brain-injured and
  normal subjects.
 Thus, most test-takers answered all items
  from Subtest I and II correctly, regardless of
  overall performance.
 Hence, performance on Subtests I and II can
  be used as measures of effort, rather than
  neurocognitive functioning.
Subscales of the Category Test

 The Category Test has received criticism for
  providing only a single Error score; although it
  requires the use of many component cognitive
 Several efforts have been made at developing
  new scales to the Category test including scales
     Categorization
     Set Maintenance
     Perseveration
     Memory
Categorization Scale

 Rosenblum, et al. (2005) developed a Category
  scale for the Booklet Category Test (BCT-CAT),
  which is generated from clusters of similar test
  items that run consecutively through each
 Moderate correlation between the BCT-CAT and
  the Categories completed scale of the WCST (r =
  .478, p < .01).
 Moderate convergent validity with stronger
  correlation between the WCST-CAT and the
  BCT-CAT than the BCT-Error score (r = .478, p =
  .01 vs. .396 p = .01).
Categorization Scale

 Discriminant validity was established against the
  WCST-Failure to Maintain Set and WCST-
  Perseveration Errors.
 The BCT-CAT also had higher correlations with
  visual and verbal memory tests than the BCT-
  Memory scales (r = .644 & .410, p = .01 vs. .408 &
  .383, p = .01).
 Land, DeFilippis, Hill, & Dodd, (2007)
  demonstrated further concurrent validity of the
  CT-CAT scales with the Halstead Impairment
Perseveration Scale

 Pelham (2001) made the first known attempts
  at developing BCT subscales intended to
  measure perseveration, loss of set, and
 Minassian, et al., (2003) published a series of
  experiments attempting to develop further
  validity and reliability for Pelham’s scales.
Perseveration Scale

 Pelham found moderate correlations with the
  WCST Perseverative Responses scale (r = .407, p
  < .01).
 Pelham also described how to score Subtest IV,
  but not Subtests III, V, and VI.
 Minassian, et al., (2003) demonstrated that the
  BCT-P correlated moderately with the WCST
  Perseverative Response scores, as well as with
  the WMS-R scores, which demonstrated good
  convergent, but poor discriminant validity of the
Loss of Set scale

 Pelham’s Loss of Set scale (BCT-LS) was
  calculated based on the total number of
  errors made after three consecutive correct
 This scoring rule was applied to subtests III,
  IV, V, and VI.
 The BCT-LS scale did not correlate with any
  other measures of attention or set
Loss of Set scale

 Minassian, et al., (2003) also were unable to
  demonstrate correlations between the BCT-
  LS scale and measures of attention and the
  WCST FTMS scale.
Memory Scale

 Pelham’s memory scale (BCT-M) is the
  percentage of possible correct scores from
  Subtest VII.
 Pelham showed strong correlations between
  the BCT-M scale and WMS-R scores.
 Minassian, et al., (2003) demonstrated a
  significant positive correlation between the
  BCT-M score and the CVLT-II learning score
  for the first 5 trials.
Scales in Development

 Revised scoring criteria for the above
  subscales has been proposed by Dodd &
 New methods and underlying constructs are
  currently in development for:
   Category scale
   Set Loss scale
   Perseveration scale
New Category Scale (CAT-2)

 The CAT-2 is established after 3 consecutive
  correct responses are achieved within a
  subtest. If the patient loses set and is unable
  to pick back up within 3 items, the CAT-2 “re-
  sets,” as it is assumed that the previous three
  items correct was by chance, and the patient
  never truly grasped the category/concept.
New Set Loss scales

 Set loss errors occur whenever a patient has
  established the CAT-2 and then responds
 2 types of set loss
   Attentional set loss
   Conceptual set loss
Attentional Set Loss (SL-A)

 The SL-A is scored only after an incorrect
  response is given after the CAT-2 is
  established and within a subgroup of items
  that have similar stimulus structure.
 These are considered “true” set loss and
  represent an unexpected change in the
  patient’s approach to the test resulting from a
  sudden dip in attention or impulsivity.
Conceptual Set Loss (SL-C)

 The SL-C is an incorrect response given after
  CAT-2 is established, between subgroups
  within a subtest in which the core stimulus
  percept changes.
 This represents weak concept formation,
  poor abstraction, poor mental/conceptual
  flexibility, and easy confusion.
New Perseveration scale

 Perseveration on the BCT occurs when a
  patient persists with the incorrect response or
  principle for two or more items despite being
  told “incorrect.”
 Further analysis of the patient’s approach to
  the test may reveal more concrete forms of
  perseveration verses conceptual forms of
New Perseveration scale

 The two most common forms of
  perseveration that are feasibly scored
  without prompting the patient include:
   counting on subtests III, V, and VI, and
   continual misplacement of roman numerals III and
    IV on Subtest IV.
New Perseveration scale

 For Subtest III, the most common form of
  PSV-W involves persisting with a counting
  principle from the previous subtest.
 PSV-W for Subtest IV manifests after the
  numbers disappear from the matrix, and the
  patient regresses back to assigning the III and
  IV to the traditional placements within the
  quadrants, rather than the clockwise
  progression of the roman numerals that is
  presented in the test.
New Perseveration scale

 Perseveration on subtests V and VI is scored
  when the patient persists with the counting
 Again, the examiner would only qualify an
  error as a perseveration if the patient persists
  with either counting objects that are the
  same (PSV- CS) or counting objects that are
  different from the other figures in that item
  (PSV- CD).
Goldstein, K. & Scheerer, M. (1941). Abstract and concrete behavior: An experimental study with special tests.
     Psychological Monographs, 53, (2, Whole Number 239).

To top