Publish the raw results. A parent selecting a school wants to ascertain what proportion of children leave it able to read and count. That is a fact about the school's output which no adjustment can improve on. Adjusted figures answer a different question -- how much the school contributed, given the population it started with -- and that is a reasonable question for a researcher. It is not the question a family is asking. Worse, the adjustment is performed by individuals who must determine what to adjust for, and every such determination incorporates a judgement about the differences between children that are the school's responsibility and the differences that are not. Those judgements are contestable and are made invisibly, inside a model the reader cannot inspect. A raw figure is at least transparent about being partial. An adjusted one conceals a whole argument inside a single number and presents it as a correction. There is furthermore a question about the intended readership of the figures. A measure designed to be fair to schools and a measure designed to inform a family are different objects. A system that publishes one and claims the authority of the other is not serving either. The honest course is to say which question the published number answers and to leave the reader to determine whether it is hers, not to answer a question she did not ask and call it assistance.
[[PASSAGE B]]
An unadjusted figure is not a measure of the school; it is predominantly a measure of who walks through the door. The correlation between intake and raw results is sufficiently high that a table of raw scores reproduces a table of the catchment. Publishing it therefore communicates to the parent almost nothing she did not know from the address, while rewarding schools for selection, not for teaching. The objection that adjustment necessitates contestable judgements is true and establishes less than it seems, because refusing to adjust is itself a judgement -- that every difference in outcome is the school's doing -- and it is the least defensible judgement available. What the objection actually supports is publication of the model, not suppression of the adjustment. Make the assumptions inspectable and the argument can be conducted in the open, the only place it can be conducted at all. And the demand for a figure that is fair to schools is not a favour to schools. A measure that penalises a school for its intake gives it an incentive to change its intake. That is the one response available to it that helps no child at all. Publishing what a school cannot change is not information; it is an instruction, and not the one anybody intends to give.
