Gifted Issues Discussion homepage Forums Identification, Testing & Assessment IQ test accuracy/ reliability

Does anyone know if iq tests produce more false positives or false negatives?

Is it possible to score high but not really be all that gifted? Ex. High achiever, practices a lot of standardized tests, sample questions, gets tutoring, etc.

Or more probable that a child is highly gifted but scores poorly on iq test due to poor test taking. Ex. Anxiety, LD masking giftedness, cultural differences, young child not mature enough, etc.

Like any measurement system, cognitive assessments (IQ tests) have standard error. This is why it is better to view scores as confidence ranges, rather than unitary measures. Typically, composite scores are best reported in 95% confidence intervals (+/- 2 SEM) (though some conventions use the 90% confidence level). If the 95% confidence interval overlaps the cut score used by some definition of giftedness, there is some expectation that this individual's nominal classification may vary by the assessment episode.

In the technical manual of a given instrument, you can also find data on reliability and validity, which gives some sense of the statistical stability of the results, and the likelihood that they measure the constructs claimed. Test publishers and designers are usually aiming for reliability coefficients of better than 0.9.

To the qualitative aspect of your question: all cognitive measures purport to assess native learning ability, rather than experience or instruction, but unavoidably confound the two to greater or lesser extents, especially with regard to language-based skills. The gold standard instruments have lesser degrees of this confound (e.g., WISC, SB, WJ, DAS, KABC). Legitimate test prep is unlikely to affect their scores by more than a few points, in the vast majority of cases. (I leave aside the issue of adults training children on copyrighted test materials, which would be blatant cheating.) I can imagine, though, that there may be a small subset of children who will have their anxiety reduced by test prep, and will thus display more of their true ability on testing. But this is not score inflation due to test prep, but score normalization, due to increased comfort level. In most cases, this kind of special prep is unnecessary, (and may even heighten anxiety for some children); assessment professionals are trained in putting children at ease, and the activities are generally experienced as fun, or at least neutral.

And yes, many more factors can result in low estimates of ability, including emotional interference, cultural/linguistic differences, 2e, and youth. In very young children, the extremely wide range of non-pathological development, in addition to the inconsistent testability of small children, makes test results less stable than later in childhood, or in adulthood.

In terms of the stability of test scores from children tested at a young age (e.g. age 6), are the scores more likely to be too high or too low?

I would say the younger you test the more likely score will be lower. My son took the wpssi at age 3 and wisc at age 8 and there was a big increase. At age three he could barely focus and sit still, let alone do the writing task.
However, I have heard of children leveling out as they get older. Ex. Child scores very high at age 5 but by age 9 they are only slightly above average.

It depends. (Typical psych response. )

On the one hand, very young children are not easy to test, which suggests that low estimates are likely. On the other hand, the difference between a highly enriched environment and an intellectually impoverished one has more impact on the performance of young children, which may result in low estimates for under stimulated children, and high estimates for those from unusually enriched environments. (Remember that norms are developed based on representative samplings, which will include some low, some high stimulation, and mostly average stimulation. So typical experience, not optimal.)

Small children are also notable for a rapid and uneven pace of development for the entire age group, across ability levels. This makes performance at one moment in time difficult to compare reliably with age peers, as the range of possible development for any one ability level is quite wide. That range also overlaps with possible development for quite a diverse range of adjacent ability levels, so that one absolute result may be attached to any one of a a huge range of possible meanings.

Anecdotally, I have definitely seen above average preschool scores move in both directions. As a statistical generalization, one would expect extremely high scores to be more likely to decrease, extremely low scores to increase, and average scores to change the least, due to regression to the mean.

Sorry that this does not really answer the question of too high or too low for your child! There are too many variables.

Same here. The psychologist who administered the test laughed at me.

We tested DS when he was mid 3s or so. I'm not really sure how accurate the test was, but it was a heck of a lot more accurate than our own parental assessment of his ability (we thought he was average or maybe high average). We've moved to a gifted private school as soon we could and it's worked out really well.

I know there isn't a simple answer... was just curious what generally was seen. DS's scores were really high, and I was wondering if they would still be that high if we retested now. To me, his scores are supported by the behaviors we see at home and others see when they interact with him, so I would expect him to test in the same range. The actual number isn't important to me, but it helps me understand how much of an outlier he is. It is really hard to wrap my head around and believe at times. The testing gives me the confidence I need to push the school to meet his needs and also to have the confidence to say that what they offer "out of the box" isn't going to be enough for him.