Since 2002, 4 million visitors plus:
hit counters
search engine optimization service

  Appletcollection Vertical Menu java applet, Copyright 2003 GD

WRAML

 

A real example of how factor analysis can change the way one understands and interprets a test is the Wide Range Assessment of Memory and Learning (WRAML; Adams & Sheslow, 1990).  On this test, 3 memory scales (Verbal, Visual, and Learning), each consisting of 3 subtests, are provided. Technical data reported in the administration manual for the test (pg 93) provide the results of a principal component analysis with a varimax rotation.  Table XXX below summarizes that data in two ways – first with the subtests grouped according to factors proposed by the WRAML authors and second by factor loading weights.

 

WRAML Scale Compositions (9 & older)

WRAML structure as proposed by authors

Verbal Scale

Visual Scale

Learning Scale

Number/Letter

(.837)

Design Memory

(.720)

Sound Symbol 

(.638)

Sentence Memory

(.749)

Picture Memory

(.674)

Verbal Learning

(.648)

Story Memory

(.196)

Finger Windows

(.584)

Visual Learning

(.401)

 

 

 

 

 

 

WRAML structure by factor analysis

Verbal Scale

Visual Scale

Learning Scale

Number/Letter

(.837)

Design Memory

(.720)

Sound Symbol

(.638)

Sentence Memory

(.749)

Picture Memory

(.674)

Verbal Learning

(.648)

Finger Windows

(.585)

Visual Learning

(.583)

Story Memory

(.695)

 

As can be seen, certain subtests do not seem to “load” well on the scales into which they have been placed. The Story Memory subtest loads on the Verbal scale with a weight of .196, while loading on the Learning Factor with a weight of .695. The “alternative” factor groupings seem to, statistically, “hang together” better than the actual test groupings.  The WRAML authors note, in the administration manual (pg 93) that “It was decided to keep the subtests in the factors which were theorized by the authors because of the logical consistency offered in Chapters 1 and 2. Further research could change this decision.”  Interestingly, Phelps (1995), in her study of the WRAML factor structure, concludes that “…data indicate that the WRAML should be revised such that the subtest placement and Index scores match empirical findings.”

 

 

Phelps, L. (1990). Exploratory Factor analysis of the WRAML with academically at-risk students. Journal of Psychoeducational Assessment, 13, 384-390.

 

To obtain a WRAML Discrepancy Calculator press here.

 

To see another aspect of the WRAML that John and I had issue with, see the Sexist Story Memory.