SkillAlign Skill Align
Australian Years 7–11 Online Practice | Year 12 Exam Packs
Cart, loading saved items

Start practising from the beginning of Year 12

Start practising from the beginning of Year 12. This independent free sample previews every topic chapter in this course before you choose a full resource. Start with topics you are learning at school, review their prerequisites and worked support, then attempt the independent questions. Use the separate solutions and feedback to plan your next step. This overview shows course breadth at representative depth; it does not demonstrate complete syllabus mastery or include the full course checks and mixed revision.

Practice Packs: Untimed, topic-by-topic practice to build understanding and skills throughout Year 12.

Exam Packs: Exam-style papers for practising under the exam or assessment conditions stated for each resource.

Our standards for 2027 Practice Packs

We develop our practice packs to help students understand the subject, assess their work and learn from mistakes. Our development and review focus on:

  • Course fit: Clear topic coverage, prerequisites and study requirements for the selected course.

  • Reasoning you can follow: Worked examples and solutions that explain methods, assumptions and conclusions.

  • Fair marking guidance: Criteria reviewed against what the task asks, including valid alternative approaches.

  • Feedback you can use: Common mistakes and practical next steps to guide further study.

  • Meaningful practice: Questions reviewed for their learning purpose and contribution across packs.

  • Readable PDFs: Notation, diagrams, page layout and question–solution consistency included in PDF review.

Each sample question below includes its worked solution and marking guidance.

SKILL ALIGN · YEAR 12 · 2027

SACE Stage 2 Mathematical Methods 2027 — Practice Pack 0 Free Sample

Questions, worked solutions and marking guidance

10 additional independent questions introduce all 10 paid-course topic chapters, with worked support and separate solutions. This representative overview shows course breadth, not full-resource depth or complete syllabus mastery.

Version 1.1 · 10 questions · 40 practice marks · Booklet build: 20261114-SMM-R03

Independent free sample. These 10 questions and their worked examples are additional to the individual and School resources. 10 additional independent questions introduce all 10 paid-course topic chapters, with worked support and separate solutions. This representative overview shows course breadth, not full-resource depth or complete syllabus mastery.

Independent Skill Align practice resource. Curriculum reference: SACE Mathematical Methods Stage 2 Subject Outline, reissued 2025; changes from 2025 onwards; all six topics (checked 26 October 2026). This resource is not endorsed by SACE Board.

How to use this resource

  1. This independent overview introduces every paid-course topic chapter at representative depth. Read each chapter's prerequisites and worked support, then attempt its questions before opening Solutions.
  2. Use the material and performance conditions stated in each question. Record topics you have not studied separately; this sample has no course checks or mixed revision.
  3. Use the worked solutions below to award one mark per criterion. Accept equivalent mathematically correct working. Carry an earlier numerical error forward where the later method is correct; do not award a mark twice for the same step.
  4. Record your reasoning, independence and next step for each topic on the review sheet. Use a full resource for further practice, course checks and mixed revision.

Untimed SACE Stage 2 Mathematical Methods learning practice across all six topics. Use approved non-CAS graphics technology where appropriate, while showing algebra, models, reasoning and contextual conclusions. Use radians in calculus. Retain full precision until final rounding; give requested exact forms, otherwise use three significant figures for approximate results. The external examination is 130 minutes, covers all six topics, supplies a formula sheet and permits two unfolded A4 sheets of handwritten notes, four sides. The official paper reference includes its Methods formula sheet on PDF page 37. School-investigation software conditions are separate from examination conditions. Sample-mean and proportion intervals use the outline's normal method with the stated sampling and approximation assumptions; an interval guides a claim rather than proving it. Different checks are not equated measures of improvement or grade predictions. Not yet studied and Not attempted must be recorded separately from incorrect work. Free Individual Pack 0 introduces all ten substantive paid-course chapters through original independent questions. Representative breadth does not replace full workshop depth, complete syllabus study or examination preparation. Supplied investigations extend familiar concepts only through the explicit rules given in their questions.

SACE formula sheet · Approved calculators

Suggested times guide study planning. Questions are grouped for learning; total marks and demand labels do not predict an external examination grade, ATAR or a validated diagnostic score.

Use the working space provided or extra paper labelled with the question reference. Keep your original attempt so you can compare its reasoning with the worked solution below each question.

DIFF · Sample workshop — reconstruct rates from unequal windows

Limited outline 1.1–1.2: average rate, a quadratic model, first-principles derivative and local lines.

Before you start: Quadratic algebra, rate units and a difference quotient; compare the actual observation windows.

Suggested study time: 25 minutes, including review.

DIFF-1 4 marks · Technology-active · Select, calculate, interpret and justify

A modelled position is s(t)=At²+Bt metres, for t≥0 in seconds. Its average velocities on [0,1] and [1,3] are exactly 3 and 5 m/s. Recover A,B and the average velocity over [0,3]. Explain why averaging 3 and 5 without interval weights is wrong. Derive s′ from the difference quotient and give the equations of the tangent and normal at t=2 in (t,s) coordinates.

Worked solution — DIFF-1

  1. The records give A+B=3 and (s(3)−s(1))/2=4A+B=5, so A=2/3 and B=7/3. The whole-window average is [3×1+5×2]/3=13/3 m/s, not the unweighted value 4.
  2. The quotient is A(2t+h)+B, giving s′=2At+B. At t=2, the point is (2,22/3) and gradient is 5. The tangent is s−22/3=5(t−2), and the coordinate normal is s−22/3=−(1/5)(t−2). The normal's reciprocal slope describes these axes rather than an additional physical velocity.

Marking guidance — one mark per criterion — DIFF-1

  • Reconstructs both model coefficients from the actual windows.
  • Calculates and explains the duration-weighted average.
  • Derives the local rate by the stated difference quotient.
  • Gives correct tangent and normal equations in (t,s) coordinates through the calculated point; no separate physical interpretation of the normal is required.

Acceptable alternatives — DIFF-1

  • Credit valid equivalent exact forms, graphs, numerical or algebraic methods with the requested evidence. Accept justified rounding from unrounded working. Follow a plausible earlier arithmetic error through a valid later method once; do not restore the earlier lost mark or award the same reasoning twice. A conclusion must fit the student's values, the original domain and the stated sampling/model assumptions. Unsupported impossible probabilities or invalid domains do not earn conclusion credit.

Common error: DIFF-1 — Averaging rates over unequal durations without weights or using an average rate as the tangent slope.

Next step: DIFF-1 — Recover the model from the complete window equations, then take the local derivative.

EXP · Exponential calculus and model decisions

Outline 1.3 and 1.5: exponential derivatives, products/composites, asymptotic behaviour, rates, extrema and curvature. Representative independent topic preview; full resources provide further depth.

Before you start: DIFF, exponential/logarithmic functions and radian trigonometry; complete stated prerequisite investigations before a mixed task, and state the model's time units and domain.

Suggested study time: 25 minutes, including review.

EXP-preview 4 marks · Technology-active · Select, calculate, interpret and justify

A fictional display is y(t)=e^t for 0≤t<ln4 and y(t)=e^t/4 for ln4≤t<2ln4. The display is deliberately reset at t=ln4. Differentiate on both open time intervals. Find its left limit and value at the reset, and assess differentiability there. Give two ordered times disproving 'positive derivative wherever defined means this complete display always increases'. Find its average rate from 0 to ln4 and explain why it can be zero despite both branch rates being positive.

Worked solution — EXP-preview

  1. The branch rates are e^t and e^t/4, both positive. At ln4 the left limit is 4 but the value and right limit are 1, so the display is discontinuous and cannot be differentiable there.
  2. For example ln2<ln4 but y(ln2)=2>y(ln4)=1. A derivative sign on separate open pieces does not remove their downward jump.
  3. The average rate is [y(ln4)−y(0)]/ln4=(1−1)/ln4=0. Net endpoint change includes the reset; it is not the total increase accumulated along the two exponential branches.

Marking guidance — one mark per criterion — EXP-preview

  • Differentiates both branches and establishes positive rates
  • Finds left limit 4, reset value 1 and non-differentiability
  • Supplies a valid ordered-time counterexample to global increase
  • Calculates average rate zero and reconciles the reset with positive branch rates

Acceptable alternatives — EXP-preview

  • Accept any two admissible ordered times with a strict decrease across the reset. Do not substitute the first branch at its excluded right endpoint or differentiate through the jump.

Common error: EXP-preview — Applying an increasing-function conclusion across an unexamined discontinuity.

Next step: EXP-preview — Check joins and actual endpoint values as well as derivative signs within each formula's domain.

TRIG · Radian calculus and changing periodic quantities

Outline 1.4 and 1.5: sine/cosine derivatives, chain factors, tangents, rates, extrema and curvature in stated domains. Representative independent topic preview; full resources provide further depth.

Before you start: Stage 1 trigonometry, radians and DIFF; account for every admissible angle/time.

Suggested study time: 25 minutes, including review.

TRIG-preview 4 marks · Technology-active · Select, calculate, interpret and justify

A phase t, measured in radians, is known only to lie within h=π/6 of m=π/3. A signal is H(t)=sin t. Find its complete output range and largest absolute departure from H(m). A technician calls |H′(m)|h a guaranteed output-error bound. Test that claim. You may use the supplied bound |H(t)−H(m)|≤h max|H′(u)|, where the maximum is over the complete phase interval. Calculate this valid bound and distinguish it from the actual largest departure.

Worked solution — TRIG-preview

  1. The phase interval is [π/6,π/2]. Since H′=cos t≥0 there, the output range is [1/2,1]. At m the output is √3/2. Downward departure is (√3−1)/2≈.366025 and upward departure is 1−√3/2≈.133975; the largest is .366025.
  2. The midpoint-rate proposal is (1/2)(π/6)=π/12≈.261799, smaller than the actual downward departure, so it is not a guarantee.
  3. The maximum absolute derivative over the interval is √3/2, giving valid bound π√3/12≈.453450. This conservative bound exceeds the actual maximum; the tangent's local sensitivity alone did not certify the whole window.

Marking guidance — one mark per criterion — TRIG-preview

  • Uses radian derivative and monotonicity to establish output range [1/2,1]
  • Compares both departures and finds the largest (√3−1)/2
  • Calculates π/12 and gives the violating endpoint
  • Applies the supplied whole-interval derivative bound and distinguishes bound from exact departure

Acceptable alternatives — TRIG-preview

  • Credit exact forms or appropriate unrounded technology comparisons. The supplied inequality may be used without proving a theorem; a valid conservative bound need not equal the actual worst departure.

Common error: TRIG-preview — Promoting a midpoint tangent estimate to an interval-wide guarantee.

Next step: TRIG-preview — Find the complete admissible interval and check the derivative's maximum on that interval.

DISCRETE · Discrete variables, Bernoulli and binomial evidence

Outline 2.1–2.3: complete probability distributions, moments, empirical evidence, Bernoulli coding and binomial models/probabilities. Representative independent topic preview; full resources provide further depth.

Before you start: Stage 1 probability and counting; distinguish the random variable from its observed value.

Suggested study time: 25 minutes, including review.

DISCRETE-preview 4 marks · Technology-active · Select, calculate, interpret and justify

Flip two independent fair coins and let X be the number of Heads. Independently choose a fair direction S, either +1 or −1, and report Y=SX. Construct Y's complete probability distribution, showing how both direction cases contribute at zero. Find E(Y) and Var(Y). A draft says multiplying by a sign preserves variance and therefore Var(Y)=Var(X)=1/2. Explain precisely why the fixed-scale variance rule does not justify that conclusion here.

Worked solution — DISCRETE-preview

  1. X has probabilities 1/4,1/2,1/4 at 0,1,2. Independent fair directions give Y=−2,−1,0,1,2 with probabilities 1/8,1/4,1/4,1/4,1/8. The two zero contributions each have probability 1/8 and combine to 1/4.
  2. Symmetry gives E(Y)=0. E(Y²)=4/8+1/4+1/4+4/8=3/2, so Var(Y)=3/2.
  3. The rule Var(aX)=a²Var(X) assumes one fixed constant a. Here S is random and changes the conditional mean from +1 to −1; the unconditional spread includes that difference. Squaring preserves Y²=X², not the original mean, so it does not preserve variance.

Marking guidance — one mark per criterion — DISCRETE-preview

  • Constructs all five masses and combines both zero cases
  • Finds E(Y)=0
  • Computes E(Y²) and variance 3/2
  • Explains the fixed-constant restriction and the changed unconditional mean

Acceptable alternatives — DISCRETE-preview

  • Accept an eight-case equally likely base-outcome table or the independent conditional calculation. A correct non-binomial signed-score law must retain its negative values and merged zero.

Common error: DISCRETE-preview — Using a constant-multiplier rule for an additional random factor.

Next step: DISCRETE-preview — Build the reported variable's full support and probabilities before calculating moments.

INTEGRAL · Sample workshop — certify an approximation before reporting it

Limited outline 3.2–3.4: monotone rectangles, a precision guarantee and an exact primitive check.

Before you start: Rectangle area, rational powers, monotonicity and integer rounding of a lower sample-count bound.

Suggested study time: 25 minutes, including review.

INTEGRAL-1 4 marks · Technology-active · Select, calculate, interpret and justify

For I=∫_0^3 √(1+x)dx, use n equal subintervals and the increasing integrand to define left and right rectangle bounds L_n,R_n. Without first using an exact integral, find the smallest n guaranteeing R_n−L_n≤0.04. At that n compute both sums and explain the error guarantee for their midpoint. Then evaluate I exactly to verify the bounds. Distinguish the guaranteed error from the actual error of this particular midpoint.

Worked solution — INTEGRAL-1

  1. The width is Δx=3/n, and all interior rectangle terms cancel in R_n−L_n, leaving (3/n)[√4−√1]=3/n. Thus n≥75, with minimum 75; n=74 fails. The resulting bounds are approximately L=4.6466333 and R=4.6866333.
  2. Their midpoint is about 4.6666333 and is guaranteed within 0.02 of I because I lies in the entire width-0.04 band. A primitive is (2/3)(1+x)^(3/2), giving I=14/3≈4.6666667. The actual midpoint error is about 0.000033332, much smaller than the certified worst-case band radius; it must not replace the prior guarantee.

Marking guidance — one mark per criterion — INTEGRAL-1

  • Defines the correct monotone rectangle bounds and telescoping width.
  • Obtains and verifies the minimum integer n.
  • Computes both sums and their guaranteed midpoint error.
  • Uses the exact primitive to compare actual and guaranteed errors.

Acceptable alternatives — INTEGRAL-1

  • Credit valid equivalent exact forms, graphs, numerical or algebraic methods with the requested evidence. Accept justified rounding from unrounded working. Follow a plausible earlier arithmetic error through a valid later method once; do not restore the earlier lost mark or award the same reasoning twice. A conclusion must fit the student's values, the original domain and the stated sampling/model assumptions. Unsupported impossible probabilities or invalid domains do not earn conclusion credit.

Common error: INTEGRAL-1 — Using a computed small actual error as a general pre-computation guarantee.

Next step: INTEGRAL-1 — First bound the whole unknown integral between rectangles, then verify the particular approximation separately.

LOG · Logarithmic models, domains and calculus

Outline 4.1–4.3: natural logs, exponential equations, inverse graphs/domains, derivatives/integrals and model decisions. Representative independent topic preview; full resources provide further depth.

Before you start: EXP, positive logarithm arguments and inverse functions; separate algebraic and contextual domains.

Suggested study time: 25 minutes, including review.

LOG-preview 4 marks · Technology-active · Select, calculate, interpret and justify

A real function F is differentiable on D=(−∞,1)∪(1,∞), with F′(x)=1/(x−1) throughout D and F(0)=0. Give the general logarithmic form on each branch. A solver reports F(2)=0 after using one integration constant. Does the supplied initial condition determine F(2)? Construct two valid functions satisfying all the given information but having different values at 2, and verify their derivatives and initial values.

Worked solution — LOG-preview

  1. On the left F(x)=ln(1−x)+C_left; on the right F(x)=ln(x−1)+C_right. Each derivative is 1/(x−1) on its own domain.
  2. F(0)=ln1+C_left=0 fixes C_left=0 but supplies no condition on C_right. There is no required continuity through x=1, which is outside D.
  3. One valid function uses C_right=0 and has F(2)=0; another uses C_right=3 and has F(2)=3, with identical left branch. Both satisfy every derivative and initial-value requirement. Thus F(2) is not determined; one common constant would impose extra information.

Marking guidance — one mark per criterion — LOG-preview

  • Gives the general logarithmic form on each stated branch
  • Uses F(0)=0 to fix only the left constant
  • Constructs two complete functions with different F(2) values
  • Verifies both functions' derivatives and initial values; the verified counterexamples justify that F(2) is not determined without requiring a separate prose explanation

Acceptable alternatives — LOG-preview

  • Accept ln|x−1| with independently stated constants on the two components. Any two distinct right constants supply a counterexample; differentiability is required only on D.

Common error: LOG-preview — Assuming one initial value determines an antiderivative across a disconnected domain.

Next step: LOG-preview — Apply each condition only on its connected branch unless a separate linking condition is supplied.

DENSITY · Continuous distributions and supported probability

Outline 5.1: valid non-normal densities, interval probabilities, moments and standard deviation. Representative independent topic preview; full resources provide further depth.

Before you start: INTEGRAL, complete support and non-negative density; probabilities are areas, not heights.

Suggested study time: 25 minutes, including review.

DENSITY-preview 4 marks · Technology-active · Select, calculate, interpret and justify

X has density f(x)=2x for 0≤x≤1 and zero elsewhere. Verify normalisation and derive its cumulative function on the full real line. A calibrated rank is U=F(X); a different display reports the density height V=f(X)=2X. Derive P(U≤u) for 0≤u≤1 and P(V≤v) for 0≤v≤2, and identify which display is uniform. Calculate P(U≤.4), P(V≤1) and E(U). Do not treat a density height as a cumulative rank.

Worked solution — DENSITY-preview

  1. The integral ∫₀¹ 2x dx=1. The cumulative function is 0 below 0, x² on [0,1] and 1 above 1.
  2. U=X², so P(U≤u)=P(X≤√u)=u for 0≤u≤1: U is uniform on [0,1]. V=2X has P(V≤v)=P(X≤v/2)=v²/4 on [0,2], which is not uniform there.
  3. Therefore P(U≤.4)=.4, P(V≤1)=1/4, and E(U)=∫₀¹ x²·2x dx=1/2. The cumulative transformation accounts for integrated mass; the density-height transformation does not do so.

Marking guidance — one mark per criterion — DENSITY-preview

  • Normalises the density and gives the full supported cumulative function
  • Derives the uniform U law by an event preimage
  • Derives the nonuniform V law with correct support
  • Calculates both requested probabilities and E(U), distinguishing height from cumulative mass

Acceptable alternatives — DENSITY-preview

  • Credit equivalent supported cumulative functions and direct event integration. No general transformation theorem or Jacobian is required; the square-root preimage suffices here.

Common error: DENSITY-preview — Using f(X) in place of F(X) when claiming uniform probability ranks.

Next step: DENSITY-preview — Translate a transformed event back to X and integrate over its supported preimage.

NORMAL · Normal models and inverse probability

Outline 5.2: density, standardisation, arbitrary intervals/tails, inverse values and model interpretation. Representative independent topic preview; full resources provide further depth.

Before you start: DENSITY, mean/standard deviation and probability; a normal model has unbounded support.

Suggested study time: 25 minutes, including review.

NORMAL-preview 4 marks · Technology-active · Select, calculate, interpret and justify

A measurement X is normal with mean 10 and SD 2. An alarm activates when X>12. A separate excess display reports E=max((X−12)/2,0) in standard-deviation units. Calculate the activation probability, the mean excess over ALL readings and the mean excess conditional on activation. You may use Z=(X−10)/2, φ(z)=exp(−z²/2)/√(2π), and the supplied identity ∫₁^∞ zφ(z) dz=φ(1). A report equates mean excess with the alarm probability because both are zero on non-activation. Audit that reasoning; no new integration technique is required.

Worked solution — NORMAL-preview

  1. Activation is Z>1, with probability p=1−Φ(1)≈.158655. The excess display is (Z−1) on that event and zero otherwise.
  2. Integrating (z−1)φ(z) over z>1 and using the supplied identity gives unconditional mean φ(1)−p≈.241971−.158655=.0833155 standard-deviation units.
  3. Conditional mean excess is .0833155/p≈.525135 units. The probability counts active readings, whereas expected excess weights each by its variable size. Multiplying the conditional mean by p recovers the unconditional mean; common zeros do not make an indicator and a magnitude the same variable.

Marking guidance — one mark per criterion — NORMAL-preview

  • Standardises the alarm threshold and calculates p≈.158655
  • Uses the supplied identity to find unconditional mean excess≈.0833155
  • Finds conditional mean excess≈.525135 and reconciles the weighting
  • Explains indicator frequency versus variable excess size with correct units

Acceptable alternatives — NORMAL-preview

  • Accept direct approved-technology integration or the supplied density identity. State whether the mean includes all readings or only active readings; no financial loss or real operational rule is implied.

Common error: NORMAL-preview — Equating an event's probability with the expectation of every quantity supported on that event.

Next step: NORMAL-preview — Write the reported variable's size on the event before integrating or conditioning.

MEAN · Sampling distributions and mean intervals

Outline 5.3 and 6.1: sample means/totals, independence, large-sample approximations, normal confidence intervals and interpretation. Representative independent topic preview; full resources provide further depth.

Before you start: NORMAL and random sampling; distinguish individual spread from standard error.

Suggested study time: 25 minutes, including review.

MEAN-preview 4 marks · Technology-active · Select, calculate, interpret and justify

Independent normal observations have unknown mean μ and known SD 10. Every sample has n=100, so the unrounded sample mean is normal. A reporting system rounds that mean to the nearest integer M (exact halves upward), then reports M±1.96, keeping the original unrounded margin. Find the standard error. For audit benchmarks μ=0 and μ=.5, list all integer M values whose reported interval covers μ, translate each set back into an event for the unrounded mean, and calculate repeated coverage. Can this rounded-centre procedure retain a uniform nominal 95% label merely because the original unrounded interval did?

Worked solution — MEAN-preview

  1. SE=10/√100=1. At μ=0, coverage requires M∈{−1,0,1}, hence −1.5≤X̄<1.5. Its probability is 2Φ(1.5)−1≈.866386.
  2. At μ=.5, coverage requires M∈{−1,0,1,2}, hence −1.5≤X̄<2.5. Centring at .5 gives −2≤Z<2, so coverage is 2Φ(2)−1≈.954500.
  3. The original unrounded procedure has coverage 2Φ(1.96)−1≈.950004. Rounding its centre while retaining the old margin changes the procedure and can make coverage depend on the true mean's position within the rounding grid. The two audit benchmarks are sufficient to disprove a uniform nominal 95% claim; they are not inferred values of an unknown μ in a real sample.

Marking guidance — one mark per criterion — MEAN-preview

  • Calculates SE=1 and uses the supplied normal sampling model correctly
  • Finds the integer centres/preimage and coverage .866386 at μ=0
  • Finds the integer centres/preimage and coverage .954500 at μ=0.5
  • Explains changed procedure and location-dependent coverage without assigning a posterior probability to μ

Acceptable alternatives — MEAN-preview

  • Credit an equivalent union of rounding bins. Open/closed endpoints do not affect the continuous probabilities. Benchmark calculations audit the procedure rather than estimating μ or proving a real-population value.

Common error: MEAN-preview — Retaining an interval label after quantising its centre without reassessing coverage.

Next step: MEAN-preview — Translate coverage through the actual rounding bins and centre at the stated audit mean.

PROP · Sample workshop — infer from a retained threshold indicator

Limited outline 5.2,6.2–6.3: binary sampling, large-count proportion intervals and inverse normal interpretation.

Before you start: Normal standardisation, Φ and its inverse; distinguish the recording parameter from the original measurement mean.

Suggested study time: 25 minutes, including review.

PROP-1 4 marks · Technology-active · Select, calculate, interpret and justify

Independent random measurements are modelled as N(μ,2²). Only whether each reading exceeds 10 is retained; 248 of 400 do. Form a 95% large-count normal confidence interval for p=P(X>10). Derive the model relation between p and μ, then map the interval into a μ confidence set using inverse normal values. Explain which conclusions remain available if the normal measurement model and known SD 2 are not justified. Do not treat the binary fraction as an observed continuous-measurement mean.

Worked solution — PROP-1

  1. The estimate is p̂=0.62 and the interval is 0.62±1.96√(0.62×0.38/400)≈[0.572432,0.667568], with counts 248 and 152. Under the extra normal model, p=1−Φ((10−μ)/2)=Φ((μ−10)/2), so μ=10+2Φ^(−1)(p), an increasing map.
  2. Mapping both bounds gives approximately [10.3651,10.8664], with transformed point estimate 10.6110. The p interval can retain its binary-sampling interpretation without the normal shape/known-SD assumption; the μ inference cannot. Its nominal coverage is inherited through that assumed model and the original proportion approximation, not proved by the threshold data alone.

Marking guidance — one mark per criterion — PROP-1

  • Calculates the original proportion interval and checks actual success/failure counts.
  • Derives the correct tail-to-mean normal model relation.
  • Inverts both probability bounds with the correct order.
  • Separates the binary inference from the extra model-dependent mean inference.

Acceptable alternatives — PROP-1

  • Credit valid equivalent exact forms, graphs, numerical or algebraic methods with the requested evidence. Accept justified rounding from unrounded working. Follow a plausible earlier arithmetic error through a valid later method once; do not restore the earlier lost mark or award the same reasoning twice. A conclusion must fit the student's values, the original domain and the stated sampling/model assumptions. Unsupported impossible probabilities or invalid domains do not earn conclusion credit.

Common error: PROP-1 — Using an indicator fraction as a continuous mean or retaining its mean inference after discarding the measurement model.

Next step: PROP-1 — Name the binary parameter first, then state every additional assumption used by its inverse map.

© Skill Align · 2027 · Practice Pack 0 · v1.1 · Questions, worked solutions and marking guidance