Why IQ testing matters - thoughts to consider...

The following comments come unedited from the OG discussion list:

From Keith McGuinness:

IQ tests provide one means of assessing "intelligence"- and potential in certain areas - that is less sensitive to training (and, therefore, is a better measure of "true" intelligence) and better standardised than other means. Whether or not such tests are "adequate" depends on the objectives of the testing.

The commonly used IQ tests (such as the WISC and Stanford-Binet, or SB) assess an individual's abilities in several domains (in the case of the SB these are verbal reasoning, quantitative reasoning, abstract/visual reasoning and short-term memory). Individuals get a score on each of several sub-scales and also have a full-scale, or composite, score. These scores are adjusted relative to the average for individuals of the appropriate age.

For the Stanford-Binet (and I think the other major tests) the average score at any age on a sub-scale is 50 with a standard deviation of 8. The average composite score is 100, with a standard deviation of 16. An individual with a composite score of 132 is, therefore, 2 standard deviations above average: scores this high or higher are expected in only about 2.3% of the population (these calculations are based on properties of the normal distribution).

The tests are standardised by being given to a large number of individuals spread over appropriate ages. The test makers attempt to match the properties of the test subjects to those in the general population, and also attempt to include representatives from various "special" populations.

An IQ test is not just a measure of mathematical and logical thinking. This is quite wrong. In fact, one common criticism, particularly of the earlier types of tests, was that the strong VERBAL component in the tests was unfair to individuals who did not speak a particular type of English as their first language. In addition to the verbal component present in some (but not all) tests there are tests of memory, and in the case of timed tests, processing speed.

The people who have constructed these tests have gone to some trouble to try to incorporate a range of items in an attempt to test different aspects of the construct we call "intelligence". To characterise IQ tests as being just tests of "mathematical and logical thinking" and reject them on this basis is incorrect.

Whether these tests adequately assess all aspects of the construct we term "intelligence" is debated, in part because researchers have not yet reached agreement on what "intelligence" is. Reference has been to the Gardner's theory of "multiple intelligences" and Spearman's "g theory". These are but two of the several alternatives which span the range from the view of "intelligence" as a single entity ("g") to that of it as consisting of numerous different "intelligences".

What is not now (seriously) debated is that IQ tests measure something which is related to potential achievement in the so-called "academic" areas. For instance, Zimbardo, PG Psychology and life (1988): "IQ scores are valid for two types of prediction: academic success and the status of one's occupation. Grades in school are significantly correlated with intelligence test scores (Tyler 1965, Wing & Wallach 1971)." and "IQ also predicts what kind of job one will obtain. Occupational status is positively correlated with level of IQ, whether status is defined as income or prestige (Brody & Brody 1976). Two limits on this relationship exist, however: (a) educational success may be the intervening state that determines the quality of the job one gets, so that IQ may really predict job status only indirectly through its correlation with academic achievement, and (b) once in the profession, intelligence score differences do not discriminate among those who are eminent and those who are less successful (Matarazzo 1972)."

And Wade, C & Tavris, C Psychology (1987): "IQ tests are well standardized and those given to school-age children predict school performance fairly well. ... Correlations between IQ scores and current or future school grades, though far from perfect, are quite high, ranging between 0.40 and 0.60 (Wing & Wallach 1971)."

Most reports indicate that training can only increase scores on IQ tests by about 5 or so points. For instance Eysench, HJ & Wilson, GD A textbook of human psychology (1976): "Furthermore, you cannot teach people how to do IQ tests; there is a minor effect from doing a few tests as practice, but that is all." And Summers, J, Borland, R & Walker, M Psychology: an introduction (1989): "One of the most intensive attempts to boost intelligence by training was conducted in Yugoslavia by Kvashchev who gave his students training each week for three years in creative problem solving. By the end of the training, the average intelligence level had risen by 8 points. After one year these gains in IQ had been maintained. The crucial question concerning whether these gains are maintained over longer periods of time remains unanswered."

From Shaun Hately:

IQ testing is far from perfect - but there is a lot of evidence to suggest it is a pretty good predictor of a person's academic success (which is a fair leader to economic success as and adult). Whether it measures intelligence or not is very debatable - but it can be considered to be a reasonably accurate measure of a limited part of human intelligence.

IQ tests are sometimes criticised as only measuring one's mathematical and logical thinking - and generally speaking that is so. That's useful - but rather limited. But other facets of human intellect correlate strongly with these two areas - for example, in very simple terms, a person who finds it easy to learn mathematics will also find it easy to learn French. A person who finds it easy to learn French will, generally, also find it easy to learn history. A person who finds it easy to learn history, will also find it easy to learn geography.

Because of the strong correlations between all of these, the IQ score, while only a measure of mathematical and logical ability, is a fairly accurate *predictor* of abilities in a wide range of areas.

An analogy - if you look at two people, one thin, muscular, and in his twenties, and the other grossly overweight, flabby, and at least 70, who would you expect to see win a tennis match between the two? The answer is obvious - you'd bet on the young thin one. Would you be right all the time? No, you wouldn't - for example, that thin, muscular person might be blind, or the 70 year old might be a former Wimbledon champion who still plays a good game. But you'd be making your best judgement - and you'd bet on the younger person, because age, fitness, and muscle tone, are all predictors of fitness, and fitness is a predictor of success in sporting endeavours. In any individual case, you can be wrong, but over a large number of cases, you will be right most of the time - and the higher the correlations the more often you will be right.

One current theory has all the various facets of intelligence (mathematical ability, language skills, social skills) all correlating very closely with a concept called 'G' (general intelligence factor), and g correlates closely with IQ. There is a lot of evidence for this theory, and it does seem IQ is a fair predictor of all types of intelligence. Not a perfect one - just a reasonable one.

IQ tests have two advantages over other forms of assessment of giftedness.

The first is very simple - IQ tests are an objective measure, rather than a subjective one - teachers' opinions of a child have virtually no effect on it, for example. This means that, while they can be wrong due to other factors, hopefully bias will not occur. And bias is a real problem - surveys of principals in poor areas, for example, have found that many report that they have no gifted children in their schools. Their bias (or more fairly their subjective impressions) are wrong - and, most significantly, they are less accurate than IQ tests. A gifted child in such an area may miss out on being identified unless anything other than an objective measure is used.

When I was in Year 7, my school *demanded* my parents get me tested because I was "retarded" and shouldn't have been in their school. I was achieving the highest marks in the Year, I was competing in public speaking competitions against kids two years older and winning, and bringing home distinctions in the science and maths competitions - but they decided I was retarded! What, in the name of reason, does a person need to do in order to get identified as gifted!?! IQ testing, flawed as it was (and the test I was given was badly administered) revealed a WISC-R score of 144 - a much better representation, for all its faults, than the school's own bias.

Mine is, perhaps, an extreme example - that school was horrid. But studies have shown that teachers can miss up to 94% of gifted children in primary classes (figures presented to the 1998 Senate Committee by the SAGTC)! Even with it's faults, an IQ test is considerably more precise than that, and misses less kids. The teacher misidentification rate isn't always that bad - training helps - but it can be pretty awful.

The other problem is that other methods of identifying gifted children tend to use achievement as a measure of giftedness. There is nothing wrong with that, in and of itself - high achievement can be a sign of giftedness, and certainly can mean a child needs access to special educational provision. But it misses one very large group - underachieving gifted children.

Again, studies have shown large numbers of gifted children underachieve - 50% in at least one study done in the US in 1983. Gifted children underachieve for many reasons, inappropriate teaching, being bored, being afraid of abuse by other kids - even pure laziness (guilty!). But any identification method with achievement as a major factor will miss a lot of these kids. And arguably, they are the ones who can least afford to be missed - those who are achieving, have at least, avoided one of the pitfalls of giftedness. An identification method that misses the people who would most benefit is pretty flawed.

 © NSWAGTC 2007

Jump to... Top | IQ testing | WPPSI R | WISC 3 | Characteristics of gifted children | Identification | Information Centre

Suggestions? | Like to Help? | Disclaimer | Webmaster | Site maintained for the NSWAGTC by Austega