Centre for Assessment
Information about assessment
Verbal reasoning tests
In the United Kingdom, group intelligence tests that are largely verbal in content and used in an educational context are often referred to as verbal reasoning tests. These tests consist of a variety of item types, typically including similes, antonyms, analogies, codes and anagrams. Modern verbal reasoning (VR) tests are objective and require a mixture of completion and multiple-choice responses, although some recent versions are entirely in multiple-choice format so that they may be machine-scored via the use of an optically-read answer sheet. VR tests are generally designed to provide an overall measure of scholastic ability without having a specific curriculum content, principally assessing inferential and deductive skills. The tests have high reliability and are relatively good predictors of subsequent academic attainment (good predictive validity).
VR tests are perhaps best seen as the inheritors of one historical tradition of intelligence assessment that predominated in the United Kingdom for many years from the early work ofSpearman (1927) through to the hierarchical model of intelligence of Vernon (1950). This can be contrasted with the American view (e.g. Guilford, 1967) that stresses the diverse nature of intelligence as a group of separate but interrelated factors. Spearman’s view of the structure of intelligence was that all test questions had a general intelligence component (‘g’) and a component specific to that question. Vernon’s hierarchical model proposed that ‘g’ could first be divided into two major group factors, a verbal-educational factor and a spatial-mechanical-practical factor. These could then be further sub-divided into minor group factors. Cattell (1963)proposed that VR tests should be classed amongst those measures that reflect ‘crystallised general intelligence’, i.e. they draw in the main on a person’s acquired knowledge and skills. This is in contrast to ‘fluid general intelligence’, which calls upon little formalised learning content but needs the ability to see complex relationships. Any one VR test cannot therefore be seen as a culture-fair test and should ideally be used only with a homogeneous population with common language and educational experience.
A criticism of this type of intelligence test is that it only provides a single global score, usually a standardised score, that gives a very limited view of a person’s ability. This can be contrasted to the much more complete view provided by a multiple abilities battery designed to assess a wider range of intelligence factors. However, VR tests are highly reliable (for tests of around 90 questions, internal consistency reliabilities are typically in the order of 0.95) and good predictors of future academic attainment (for example, correlations with school exams taken over five years later are above 0.70).
For this reason they were commonly viewed as indicating capacity to learn and were widely used in the United Kingdom as part of the process for selecting children at the age of 11 for a separate academic education in grammar schools (the 11+ test). Whilst selection for entry to grammar schools still continues in some areas in England, the view of verbal reasoning tests as fixed measures of potential has declined and they are now better regarded as measures of reasoning which reflect a person’s educational experiences up to the time of testing. (Whetton, 1985).
The following are relatively straightforward examples of verbal reasoning questions:
Underline two words, one from each set, that are closest in meaning:
Set 1. office shop start Set 2. work begin end
Re-arrange the letters in capitals to make a correctly spelt word that will fit into the sentence:
He tried his best to catch the LABL
If the code for TEAR is Q34M, what is the code for ATE?