Take three or four hands-up guesses and write them on the board, but don't rank them yet, just collect them. Give five seconds of quiet think-time before any hands go up.

The spread of guesses is the hook: some will say 234, some 240, some 230 exactly. That range of answers is what the lesson will explain.

Work each method live, don't read answers off. Mean first, let it settle: 210 + 216 + 220 + 226 + 230 = 1102, ÷ 5 ≈ 220. Point at the dashed line — it sits below Year 5 (230). Ask why: draw out that the older, smaller years pull the average down, so the mean under-predicts a rising trend.

Step method: read the jumps aloud, +6, +4, +6, +4, about +4 or +5 a year. So 230 + 4 = 234 — the first new bar on the second chart.

Contrast so far: mean says about 220, step says about 234, same five years.

What-if: present as a tweak to the step, not a new method. "What if a new estate opens and adds 30 pupils?" 234 + 30 ≈ 264 — the tallest bar, second new one.

Try Together — talk this through as a class on the board; nobody is marked and nothing is saved. Use a pacing lever between each placement so the watching class stays with it.

Enter the five bars with the class. Before each pupil drags the sixth bar, ask the pupil to name their method aloud, then ask the whole class "where will this land?" and take two quick answers before the bar moves. After it lands, revoice what the pupil did.

Rotate three pupils so the class sees the mean method (~220), the step method (~234) and a what-if (higher) as three different sixth bars. Close by revoicing: "so the same five years give us three different next years, depending on how we read them."

This is a paper / seat task worked in the copy, not on screen. The task bank below renders as a visible list. If you printed the prediction-table worksheet, hand it out now; otherwise pupils rule the three-column table into their copies from a model on the board. Circulate while pupils work, and check each pupil has named a method and is showing steps rather than guessing.

• Ballybane rises by a steady +8: mean = 980 ÷ 5 = 196, step method 212 + 8 = 220, what-if 220 − 20 = 200. Watch for pupils who forget to subtract the 20 in the what-if.
• Douglas hovers near 300: the step method is unreliable here (the jumps flip sign), so the mean (1,496 ÷ 5 = 299.2 ≈ 300) is the confident call. That contrast is the point of the second school, so weaker pupils can do just the mean and the judgement.

The stretch is a judgement call, not a calculation. Accept any answer that names a method and gives a reason.

Paper journal, not typed, not marked right-or-wrong. Model the full write-up on the board with Ballybane first so every pupil sees three numbers, a choice, and a reason before they write.

Board model (Ballybane): mean = 980 ÷ 5 = 196; step = 212 + 8 = 220; what-if = 220 − 20 = 200. Trusted: step 220, because the climb is steady +8 each year.

Douglas fix: no what-if was set earlier. Give the one in the board text now (15 join). Pupils start from the mean (~300) and add 15 → about 315, since the step was judged unreliable on the wobble. Slip to watch: Douglas pupils still forcing a step number, or reasons that name no method.

Walk the room. Glance that each entry names a method and ties the reason to the shape of the data. A Douglas pupil who says "the mean, because the numbers wobble" has understood the whole lesson.