If the member of staff who is doing the assessment knows whether pupils are in the intervention or control group bias can creep in. This is not about cheating or lying – the bias is subconscious and inevitable however honest we think we are being. There are two main ways to address this:
Eliminate any judgement from the assessment: Objectively marked items (such as multiple choice tests) do not require judgement, so are less likely to be prone to any expectation effects. You could buy in online or digital tests that include automated marking. However this is not always possible.
Ensure the testing is done ‘blind’: Where any judgement is involved the outcome assessment should be done by a member of staff who does not know which pupils are which, or the papers should be marked anonymously. This is sometimes called blinding. For example, if essays or short answers are to be marked, candidates’ names should be concealed. If a teacher might recognise handwriting or other features, then someone who does not know those students should mark it. If different classes (or schools) received different interventions then their scripts should be mixed up before marking.
This may seem like a lot of trouble to go to and in some cases may not be possible. However, there is a lot of evidence that un-blinded judgement-based assessments are biased, often substantially so.