Cognitive styles are defined as an individual's preferred or habitual way of processing information. In this thesis,
existing measures of cognitive style are reviewed and the development of a new more reliable and valid cognitive
style measure is outlined. The research described in this thesis includes (i) an introduction to cognitive styles research,
(ii) an evaluation of Riding's (1991; 1998) popular Cognitive Style Analysis (CSA) test, (iii) the design and pilot of an
inspection-time test of wholistic-analytic style, iv) the development oftwo new reliable tests of verbal-imagery and
wholistic-analytic cognitive style (VICS test, and Extended WA test) and (v) the validity of the new tests and their
potential impact on theory and practice.
Many different cognitive style measures have been proposed. The most popular computerised cognitive style test in
the UK is Riding's Cognitive Style Analysis (CSA) test. The CSA assesses style on two broad dimensions: verbalimagery (VI) and wholistic-analytic (WA). This thesis examines the test re-test, parallel form and split-halfreliability
of the CSA's VI and WA dimensions in 50 participants. The results showed that the CSA's crucial style ratios, which
are used to determine a person's cognitive style, were not reliable (r = .20 on the VI dimension, and r = .30 on the WA
To try to improve the CSA's test of the wholistic-analytic dimension, an inspection time test of the WA dimension
was designed and piloted. The inspection time WA ratio was found to be moderately reliable over time (r = .595). This
line of investigation was put on hold, with the discovery that extending the length of the WA tests increased the splithalfreliability of the Extended WA test ratio (mean r = .67). The split-halfreliability of the Extended VI test ratio did
not improve (mean r = .36). These findings suggested the need to revise the CSA's test of the VI dimension.
Therefore, a new computerised test of verbal-imagery cognitive style (VICS test) was designed and tested. The crucial
verbal-imagery ratios from the new VICS test were found to have test re-test reliability in 50 subjects of r = .66, and a
split-half reliability of r = .72, whereas the reliability of the VI ratios from CSA remained low (r = .32).
Finally, the validity of the new Extended WA and VICS tests was examined in 100 participants; specifically, the
relationship that the VICS and the Extended WA had with personality (measured with IPIP, EPQ-R, IVE), intelligence
(measured by 8 tests from the Kit of Factor Referenced Cognitive Tests) learning style (measured by ASSIST, PEPS)
and behaviour (measured by the Instructional Preference Questionnaire and observed behaviour). No personality trait,
cognitive ability, learning style or behaviour measure correlated more than .33 with cognitive style. It is suggested that
cognitive style, as measured by the Extended WA test and the VICS test, is independent from personality, ability and
2 other learning style measures. The potential impact of these tests on current theory and practice is discussed.