Whitaker, Simon (2008) Can we measure low IQ. In: British Psychological Society North East Branch (NEEB) Conference 2008, Thursday 26th - Friday 27th June 2008, Leeds, UK. (Unpublished)
Abstract

Introduction
It is usually considered that the error in measuring FS IQ at all levels of ability is about five points either side of the obtained score. However, this only takes into account the internal consistency of the test; it does not take into account other sources of error. These errors were analysed.

Method
The following studies were completed to investigate this: A meta analysis of the stability of low IQ (<80), an analysis of the a the floor effect in the commonly used IQ tests the WISC-III and WAIS-III, a comparison between the WISC-IV and WAIS-III on 16-year-olds.

Results/outcome
The mean stability of low IQ was found to be .82 which equates to a 95% confidence interval of 12.5 points either side of the obtained score. This error should be added to that due to lack of internal consistency. Error due to an unacknowledged floor effect, resulting from scaled scores of one being given for very low raw scores or raw scores of zero could increase scores by one or two points. It was found that the WISC-IV always produced a lower score than the WAIS-III, the average difference being just less than 12 points.

Conclusion/Discussion
The above results, together with other recent results in the literature on the Flynn Effect (the systematic change in intellectual ability for the population as whole form one generation to another) means that when all the sources of error a added the following confidence interval emerge for the WISC-IV and WAIS-III. For the WISC-IV there is a chance error of 15 points, to which must be added two points due to uncertainty as to how the Flynn Effect has affected the intellectual ability of people with low IQs since the WISC-IV was standardised which gives an effective 95% confidence interval of 17 points. It may also measure 10 points too low due to other systematic errors demonstrated by the difference with WAIS-III but possibly measure one or two points too high due to the floor effect, suggesting that in addition to the confidence interval it may measure eight points too low. The effective confidence interval extends 25 points above the measured IQ and 17 points below. For the WAIS-III there is a chance error of 15 points, to which must be added three points due to uncertainty as to the degree to which the Flynn Effect has effected low IQs since the WAIS-III was standardised giving an effective 95% confidence interval of 18 points. It may also measure 10 points too high due other systematic error demonstrated by difference with WISC-IV and possibly measure one point to high due to the floor effect, suggesting that in addition to the confidence interval it may measure 11 points too high. If these sources of error are added together then the effective confidence interval extents 18 points above the measured IQ and 29 points below.

Library
Statistics
Add to AnyAdd to TwitterAdd to FacebookAdd to LinkedinAdd to PinterestAdd to Email