CAPÍTULO IV: RESULTADOS
4.7. FRECUENCIAS IDENTIFICADAS DE ENGRANAJES
4.7.1. Estación de engranajes fijos helicoidales
The General Reasoning Test Battery (GRT2) was designed to serve as a composite psychometric measure of intelligence within the workplace (Budd, 1993). The GRT2 comprises the commonly tested and important verbal, numerical, and abstract reasoning ability subtests (Ackerman, Kanfer, & Goff, 1995; Ackerman & Rolfhus, 1999; Heim, 1970).
The GRT2’s composition is consistent with this investigation’s Cattell-Horn-Carroll (CHC) conceptualisation of intelligence. Verbal, numerical, and abstract reasoning represent first stratum components of the CHC model and are subcomponents of the second-stratum factors of fluid and crystallised intelligence. Fluid and crystallised intelligence are in turn subcomponents of stratum- three’s general intelligence factor (Carroll, 1993). As well as functioning as measures of the second- stratum factors of fluid and crystallised intelligence, the GRT2 components of verbal, numerical, and abstract reasoning can be summed to provide a general intelligence score (Cowan, 2005; Jensen, 1998b).
The verbal reasoning component assesses one’s knowledge of words and ability to use words in a rational way. This latter requirement involves accurately identifying logical relationships between
87 words and using these relationships to draw appropriate inferences and conclusions (Budd, 1993). This verbal reasoning subtest primarily loads on crystallised intelligence (Carroll, 1993).
The abstract reasoning component assesses the ability to identify logical relationships between abstract geometric patterns and spatial relationships. This type of abstract reasoning assesses fluid intelligence. This means the GRT2’s abstract reasoning component loads upon reasoning ability that is not greatly affected by educational and cultural experience (Budd, 1993).
The numerical reasoning component assesses the capacity to use numbers in a reasonable way, correctly identifying logical relationships between numbers and drawing appropriate conclusions and inferences upon this basis. The numerical reasoning component of these measures is less clear-cut than verbal or abstract reasoning in its loading on crystallised or fluid intelligence. Numerical reasoning assessments are generally comprised of relatively novel and thus reasonable assessors of fluid intelligence. Yet a certain minimal level of arithmetic attainment (Gc) is required because numbers cannot be manipulated except by mathematic rules (Kline, 2000). A basic level of arithmetic attainment is also required to accurately answer GRT2 numerical reasoning items (Budd, 1993).
General Properties
The GRT2 is a power test targeted at the general population. The technical manual suggests no significant difference between versions administered in paper and pencil or computer format, or scored manually or electronically (Budd, 1993). This is consistent with other findings concerning the
psychometric equivalence of computer and booklet forms of tests (Finger & Ones, 1999). Those undertaking the GRT2 are asked to work as quickly and accurately as possible and are given the opportunity to try example questions prior to undertaking the test proper.
The GRT2’s verbal component has 35 items with six response options and has an eight minute time limit. An example item follows:
Sick means the same as?
88 The numerical component has 25 items with six response options and a time limit of 10 minutes. An example item follows:
The abstract component has 25 items with six response options and a time limit of 10 minutes. An example item follows:
Psychometric Properties
Information pertaining to the reliability and validity of the GRT2 follows. This information provides support for this measure’s capacity to assess verbal, numerical, and abstract reasoning ability.
Reliability Evidence
A test-retest investigation after a two week interval carried out by the test supplier found acceptable levels of consistency across testing occasions: verbal (r = .81), numerical (r = .84), and abstract (r = .78) (N = 54). The GRT2 technical manual also details Alpha coefficients for its three subscales as: verbal (r = .83), numerical (r = .84), and abstract reasoning (r = .83) (N= 135). That each of these reliability coefficients exceeds 0.8 supports the GRT2’s internal consistency (Budd, 1993; Psytech, 2009).
Validity Evidence
Construct validity support provided by the supplier for the GRT2 comes from comparisons with other ability assessments. The first reports correlations between GRT2 and The Group Test of General
Which of the following is the odd one out?
1. 2/4 2. 6/8 3. 3/6 4. 4/9 5. 4/8 6. 2/6
89 Reasoning Skills (AH3) subscales, and total scale scores for both measures (N = 81). The AH3 is designed to measure the cognitive ability of the general population (Heim, Watt, & Simmonds, 1974). Correlations were as follows: verbal (r = .63), numerical (r = .76), abstract (r = .56), and total (r = .82) (Budd, 1993; Psytech, 2009). The strength of these correlations indicates that the GRT2 and AH3 are measuring similar constructs. Yet correlations between the GRT2 and AH3 fail to provide clear support for the discriminant validity of GRT2 inferences. This is because the correlations between each of the GRT2 subscales and their respective AH3 counterparts (e.g., GRT2 Verbal with the AH3 Verbal) are not significantly greater than correlations across other presumably less related subscales (e.g., GRT2 Abstract with AH3 Verbal [r = .61]).
The GRT2 technical manual also details correlations between the GRT2 and other measures of comparable constructs, namely the Technical Test Battery (TTB2) and the Clerical Test Battery (CTB2). General Reasoning Test subscales have modest correlations with the TTB2 (N = 94) ranging from .40 to .47. When the GRT2’s verbal reasoning subtest was compared to the CTB2 (N = 54) the strongest correlation was not as expected with the Spelling measure (SP2) (r = .34), but instead with the Office Arithmetic scale (NA2) (r = .51). Nevertheless, subsequent examination of the NA2 does reveal a reasonably high verbal problem-solving element, which may explain this relationship and provides further support for the validity of inferences based upon GRT2 scores.
A number of criterion-related validity studies for the GRT2 are also reported. In a concurrent investigation the GRT2 significantly correlated with the numerical (NR2, r = .29; AR2, r = .31) and software (NR2, r = .32; AR2, r = .28) related skill areas of 118 retail bankers. In another study involving service engineers (N = 46), rated performance was found to correlate strongly with the GRT2 subscale of verbal reasoning (r = .46), and moderately with the abstract reasoning subscale (r = .28) (Budd, 1993). Another study found the job performance of experienced newspaper printers (N = 70) was correlated with abstract (r = .36), verbal (r = .26), and numerical reasoning (r = .28). A further investigation examining the aggregate examination performance of 100 trainee Financial Consultants found correlations with numerical (r = .46) and abstract reasoning (r = .44). These
90 investigations and others provide evidence for criterion-related validity in support of GRT2 inferences (Budd, 1993; Psytech, 2009).