In his letter of February 13, Professor Keith Topping states that he used free school meal (FSM) uptake figures in his analysis of exam results. In his letter in the same issue, Ron Mitchell makes a convincing case that these were entitlement figures.
We must ask, however, why schools serving socially deprived housing estates in Aberdeen have entitlement figures so much lower than Glasgow. Is unemployment much lower in Aberdeen? If so, this may point up another problem with FSM - it will shift with temporary changes in employment patterns and not adequately reflect the underlying socio-educational nature of the pupils' backgrounds.
Mr Topping, in defending his use of FSM and asking "what other index might be better?", misses the main point. FSM might be the best available deprivation measure but no deprivation measure, used alone, can give us reliable information on school effectiveness.
To get this, we must measure and statistically control all the factors other than school effectiveness which significantly influence exam results.
In addition to a deprivation measure an advantage measure, such as percentage of parents with post-compulsory education, is essential.
The inspectorate provides each secondary school with information about how its pupil attainment compares with 20 "similar" schools, using only FSM to define similarity. This information is thus highly misleading and must favour schools with a high percentage of advantaged pupils for that "similarity" grouping. To say this is just one indicator among many will not do: the public will take these invalid comparisons seriously.
To the inadequacies of other FSM methods, Mr Topping adds the absurdity of a completely arbitrary "compensating" formula. Because schools with a very low FSM can't possibly reach the adjusted scores of the best 50, he admits to "some mathematical anomalies". He does not seem to realise that his method renders all his results uninterpretable and gives us no reliable grounds for picking out those schools which are "worthy of more detailed study" because they are performing "beyond expectations".
We know that "raw" tables are unfair but alternatives like Mr Topping's, which break the most basic rules of statistical inferencing, are not helpful.
Michael Davenport Bank Road, East Linton