A fictional 11-16 school of about 900 pupils. Referrals for "low-level disruption" rose from 260 to 412 between the autumn and spring terms, and senior staff have asked for research. In January the behaviour policy changed: referral, previously discretionary, became the expected response to a second warning.
So nobody yet knows whether pupils are behaving differently. The log records staff referrals, not pupil behaviour, and the thing that most obviously changed is the rule about when to refer. The school's full evidence file sits inside the activity below, and it is the same file for every lens.
All figures are invented for teaching. They are not estimates of anything, and no real school, policy or dataset is described.
Three research teams are given exactly the material above. They differ only in what they assume the social world is like and how it can be known.
Declare a starting point, then make five decisions. Each decision offers the same four options whichever lens you declared - the tool never labels an option with the position it belongs to, because in a real research meeting nobody does either.
Nothing here changes the controls for you, and none of it says what you are going to find.
The specification on the right fills in as you choose. Watch what happens to the rows you have not touched.
Case: a reported rise in low-level disruption at Meadowbank Academy.
| Starting point | Object of study | Becomes visible | Recedes from view | Action it points to |
|---|
Each sentence below could appear in the school's report to governors. For each, decide which design - if any - could support it.
Methods are usually taught as a toolbox: pick the technique that suits the question. That hides the step before it. All three teams held the same log, the same survey and the same policy change; what differed was what each took disruption to be, and that decision settled the question, the evidence and the claim before anything was collected. Choosing a method looks like the decision. It is the consequence of one already taken.