Most students who fail Cambridge A-Level Physics do not fail because they cannot memorise formulas. They fail because the exam asks them to decide — which model applies, which assumption is safe, what the uncertainty in a measurement is, and whether a result is physically sensible. That decision layer is invisible in the syllabus and is exactly where the marks live.
This is why Cambridge A-Level Physics (9702) rewards a different kind of preparation than most school subjects. You can know every equation in mechanics and still lose half the paper on a single structured question about a falling object, because you never asked whether air resistance was negligible. The subject trains a habit: translate a messy physical situation into a clean model, compute, then sanity-check the answer against reality.
Below is a practitioner's view — how the reasoning actually works, where the scheme of the exam breaks down, and which learning paths genuinely fit.
What the subject really is: modelling, not formula recall
Physics at this level is a discipline of controlled approximation. Every worked problem hides a modelling step: treat the string as massless, the field as uniform, the gas as ideal, the collision as elastic. The skill being tested is whether you can pick the right approximation and defend it.
That is why the AS and A2 halves of the course feel different. AS (mechanics, waves, matter, electricity basics) rewards careful bookkeeping — units, significant figures, vector signs. A2 (fields, oscillations, thermodynamics, quantum and nuclear physics, medical and astronomy contexts) rewards transfer: the same energy argument shows up in a capacitor, a gravitational orbit, and a photon. Students who learn topics as separate boxes hit a wall here, because the exam deliberately reuses the same reasoning in unfamiliar clothes.
One walkthrough: a projectile with a twist
Take a classic structured question. A ball is launched at 20 m/s at 35° above the horizontal from a cliff 15 m high. Find the horizontal distance travelled before it lands, and state one assumption.
The naive move is to plug into a range formula and stop. That is where marks disappear. Here is the order a competent solver uses:
- Resolve the initial velocity. Horizontal component = 20·cos35° ≈ 16.4 m/s; vertical = 20·sin35° ≈ 11.5 m/s upward.
- Choose a sign convention and stick to it. Take upward as positive. The ball starts at +15 m and lands at 0 m, so displacement is −15 m.
- Solve the vertical motion for time. Using s = ut + ½at² with a = −9.81: −15 = 11.5t − 4.905t². Rearranged: 4.905t² − 11.5t − 15 = 0. The positive root is about 3.16 s (the other root is negative and physically meaningless here).
- Get horizontal distance. 16.4 × 3.16 ≈ 51.8 m.
- State the assumption. Air resistance is negligible, so horizontal velocity stays constant.
Now look at what breaks if you do it differently. If you forget the cliff height and treat it as ground-to-ground, you get roughly 2.35 s and about 38 m — a 25% error that a careful marker spots instantly. If you keep the negative root, you report a negative time. If you round 16.4 to 16 too early, your final answer drifts and the significant-figures mark goes with it.
That structure — resolve, fix signs, solve the constrained direction first, then the free direction, then justify the model — is the same skeleton for a capacitor discharge, a circular orbit, or a photon-energy question. Learning the skeleton is worth more than learning forty formulas.
Where the scheme itself breaks
The exam's logic has failure modes — not student mistakes, but places where the standard approach stops working. Spotting them early is a skill in itself.
- The approximation becomes invalid. A projectile model without air resistance is fine for a dense ball at moderate speed, useless for a feather or a high-speed object. What goes wrong: you compute a clean answer that the question's own context contradicts. Early signal: the problem mentions a light object, high speed, or a long fall. If it is already too late, state the limitation explicitly — examiners often award the reasoning mark for recognising it.
- The data contradicts the model. In practical work (P3 and P5), a graph that should be a straight line bends at the ends. What goes wrong: students force a line of best fit through a curve and report a gradient that means nothing. Early signal: residuals cluster at the extremes. The fix is to restrict the range, explain the curvature physically, and report the uncertainty honestly.
- The instrument's resolution is the real limit. If a stopwatch reads to 0.01 s but human reaction time is around 0.2 s, quoting a time to two decimal places is fiction. What goes wrong: uncertainty analysis becomes decorative. Early signal: the quoted precision is finer than the apparatus can deliver. The remedy is to identify the dominant uncertainty and propagate that, not the nominal one.
- The planning question has no single right answer. In P5, you design an experiment; there is no template. What goes wrong: students write a method that cannot actually test the relationship asked about. Early signal: your variables do not map onto the equation you are testing. The fix is to start from the equation, identify what you vary and what you measure, and only then write the steps.
Learning paths and their real trade-offs
There are three honest routes to competence, and they cost different things.
| Path | Time | Money | What you actually get |
|---|---|---|---|
| Self-study from a textbook | High, self-paced | Low | Full coverage, but no external check on your reasoning; you cannot tell a good assumption from a plausible one |
| School or tutor | Structured | High | Feedback and pacing, but depends heavily on the individual teacher's depth |
| Structured online course | Medium | Moderate | A sequence that follows the syllabus, with feedback on your decisions where you would otherwise guess |
The trade-off is uncomfortable and worth naming. Self-study is cheapest but the failure mode is silent: you practise wrong reasoning for months and only discover it in the exam. A tutor fixes that but is the most expensive per hour and quality varies enormously. A structured course sits in between — it gives sequence and a place to check your thinking, but it cannot replace the discipline of doing past papers under time pressure.
What actually moves the needle in any path is feedback on decisions, not more content. You already have the content. The scarce resource is someone — or something — telling you that your assumption was wrong before the exam does.
Who this path suits — and who it does not
Cambridge A-Level Physics (9702) suits students who intend to study engineering, physics, medicine, or another quantitative science, and who are willing to reason rather than memorise. It also suits anyone who enjoys the moment a messy problem collapses into a clean model.
It does not suit students looking for a light qualification, or those who need fast visible progress to stay motivated — the payoff here is delayed and cumulative. If you are choosing between subjects and want breadth, a related quantitative course may serve you better; the reasoning habits overlap more than the topics do, as the Cambridge International A-Level Chemistry (9701) guide shows from the other side of the same skill.
If your weak point is the study discipline itself rather than the physics, the cognitive-science view of Procrastination is a useful complement — physics punishes inconsistent practice more than almost any subject.
A step-by-step plan that actually holds
Start with AS mechanics and waves, because they teach the modelling habit in the most concrete setting. Then move to fields and oscillations, where the same energy and force arguments reappear in abstract form. Leave quantum and nuclear until the reasoning is automatic, and treat the practical papers as a separate skill to be drilled, not absorbed.
At every stage, the loop is the same: attempt a structured question, write the reasoning, compare against the scheme, and note the assumption you missed. The Cambridge International A-Level Physics (9702) course follows this sequence and gives you a place to check your decisions — including feedback on written reasoning through the course chat — so the silent-failure problem of self-study is closed. The lessons adapt to your current level, and the course walks through exactly the kind of modelling steps described above with worked examples.
Begin with one past paper, one honest attempt, and one list of the assumptions you missed. That list is your real syllabus — everything else follows from it.