Due to the relative stability of measures of ability, interest, and personality, these three areas are the
primary focus of people using psychological tests to predict future behavior.
One important assumption that is made when measuring these attributes is that :
This assumption of normal distribution is necessary in order to analyze data with parametric statistics.
When using psychological attributes in making predictions, it is absolutely vital to match the appropriate
attribute with the predicted behavior.
For example, measures of general intelligence wouldnt be of much use in determining an appropriate
psychological treatment.
What is Intelligence ?
Intelligence is a construct, not a concrete object.
Researchers disagree on what the definition of intelligence should be.
Although we may not be able to definitively answer what intelligence is, we can list behaviors which we
feel represent some level of intelligence:
Being able to solve a Rubiks cube without physically taking it apart and reassembling it.
Additionally, none of these behaviors alone can be said to represent the entire range of intelligent
behavior.
Scientists believe that intelligence is a valid and useful construct for two reasons :
First, a wide variety of mental processing tasks show systematic individual variation. An individual who
performs well on one measure of cognitive ability, will likely perform well on other measures of mental
processing.
Second, this construct is related to success in a wide variety of life tasks : school performance, training
programs, and work behaviors.
So, since intelligence does exhibit consistent individual differences and can be used to predict
performance in a number of important areas, Psychologists study general mental ability extensively.
General Mental Ability (intelligence) : The performance of tasks involving the manipulation, retrieval,
evaluation , and/or processing of information which shows individual differences.
3. An accurate measure of g should reliably predict performance on a wide range of mental tasks.
Fluid Intelligence : the ability to see relationships, i.e. analogies and number and digit series completion.
For example : 2 4 8 16 ___ ___ ___
Carrols (1993) three levels of cognitive ability model follows the same principle. However, Carrols
model was developed with the aid of factor-analysis, a statistical technique which clusters responses on
a particular dimension.
Carrols general factor g is broken up into 7 classes of broad abilities:
Fluid Intelligence
Crystallized Intelligence
General Memory
Visual Perception
Auditory Perception
Retrieval Ability
Cognitive Speediness
These broad abilities can be broken up into specific abilities
Hierarchical Models suggest that both broad and general abilities can be measured.
The key is to determine what you are trying to predict, and to match the type of intelligence which is
most relevant to making your predictions.
If you wanted a general classification of children, for example, than a standard measure of g would be
highly appropriate.
If you wanted to be able to distinguish among different levels of mechanical abilities, then a narrowly
focused ability test would be more appropriate.
Rejecting the idea of g, Guilford has theorized that intelligence is best seen as a function of Content,
Operations, and Products in a three dimensional model.
Guilford identified 6 types of specific operations,
5 types of specific content, and 6 types of products produced by thought. All together this represents
180 different types of specific intelligence.
Cognitive theories like Sternberg's will have profound impact on the next generation of intelligence
tests.
1. Broad sampling of tasks : If you believe that there are 7 (or 180) Broad classes of mental ability, then
you should have problems which address each type of mental category.
2. Sufficient sample of items within task type. 30 or more would be statistically ideal, less than 10
would be statistically shaky.
3. General Intelligence tests should not test specific content. Achievement type items, which show
mastery of specific subject area, should be kept to an absolute minimum. Remember, we want to tap
into the process of intelligent action, not retrieval of specific facts.
4. Indifference of the indicator: underlining the point made in three, the specific content of a test item
is not nearly as important as the content of the underlying process which produces a specific answer.