3.8. Two aspects of statistics that are useful in analysing their quality are their bias and dispersion.
3.9. Bias is a measure of the extent to which initial estimates are lower or higher than the final estimate and thus gives an indication of the direction of revisions. In this study it is calculated as the average of the differences between the initial estimate for each period and the latest estimate published for the same period, with positive and negative revisions being netted against each other. This measure can therefore be described as the average of the values of all revisions taking account of sign.
3.10. Dispersion is a measure of the spread of final estimates and gives an indication of the magnitude of revisions. It is calculated as the average difference between the initial estimate for each period and the latest estimate published for the same period, but with positive and negative revisions not being netted against each other. It can therefore be described as the average of the absolute values of all revisions, without regard to sign.
3.11. In this study, the measure of the average used is the median value; that is the value of the middle observation in a group of observations ranked by value. This measure is preferred to the mean value (derived by summing all observations and dividing by the number of observations), as an expression of central tendency because, in these revision statistics, there are significant outliers that exert a disproportionate influence on the mean value.
Print Page
Print All