Sunday, 17 February 2013
The Bell Curve and Standarised Scores
In order to assert clinical opinion toward a patients raw data test results from varying sources should be assimilated into a common metric (standardisation). Percentiles tell us the rarity or abnormality of an individual's score. They are easily understandable and communicable but are not linear and so the difference between the 10-20th percentile in comparison to the 20-30th percentile is not necessarily the same. Z scores on the other hand, are linear transformations. They indicate with positive and negative values how many standard deviations a score is away from its mean. However, working in negative can present communication problems and z scores tend to get lost in translation when communicating with non-neuropsychologists. A common alternative is to use T scores (mean=50, SD=10) which offers a balanced level of incrementation. Others prefer index scales (mean 100, SD 15) which are often received with familiarity as they are used in the measurement if IQ. However, the fact that they are used in IQ can bring unwanted misconception when communicating interpretation and opinion.