Vertex Learn

2026 NCEA AS 91267 exam answers

Question ONE

Question ONE

a(i)
Achieved

Advert tree: P(teen) = 0.5, P(young adult) = 0.35, rest adults. P(watch | teen) = 45, P(watch | young adult) = 0.45, P(not watch | adult) = 0.85. Complete the tree and find P(young adult and does not watch).

Answer

0.1925

Step by step
  1. First fill in the missing branches: each set of branches from one point adds to 1.
  2. Adults: 1 − 0.5 − 0.35 = 0.15. Teen not watching: 1 − 0.8 = 0.2. Young adult not watching: 1 − 0.45 = 0.55. Adult watching: 1 − 0.85 = 0.15.
  3. Multiply along the young adult, Not W path.
Full working
  1. Completed tree: P(Adult) = 1 − 0.5 − 0.35 = 0.15; Teenager: W 0.8, Not W 0.2; Young adult: W 0.45, Not W 0.55; Adult: W 0.15, Not W 0.85.
  2. P(YA and Not W) = 0.35 × 0.55multiply along the branch
  3. = 0.1925
a(ii)
Achieved

Using the advert tree, find the probability a random user watches the advert.

Answer

0.58

Step by step
  1. Find each path that ends in W and add them.
  2. Teen and W, young adult and W, adult and W.
Full working
  1. P(W) = 0.5 × 0.8 + 0.35 × 0.45 + 0.15 × 0.15add the three W paths
  2. = 0.4 + 0.1575 + 0.0225
  3. = 0.58
a(iii)
Merit

In a group of 410 users, how many would you expect to be teenagers or to watch the advert?

Answer

About 279 users (278.8)

Step by step
  1. 'Teenager or watches' includes every teenager, plus the young adults and adults who watch.
  2. Do not count teenagers who watch twice.
  3. Multiply the probability by 410.
Full working
  1. P(T or W) = P(T) + P(W) − P(T and W) = 0.5 + 0.58 − 0.4addition rule
  2. = 0.68check: 0.5 + 0.1575 + 0.0225 = 0.68
  3. Expected number = 0.68 × 410 = 278.8
  4. About 279 users
b
Merit

If only teenagers and young adults were targeted, would that reach at least 96% of all users who watch the advert? Justify using the tree.

Answer

Yes, just. Teenagers and young adults make up 0.55750.58 ≈ 96.1% of all watchers, which is at least 96%.

Step by step
  1. We need the share of all watchers who are teenagers or young adults, a conditional probability.
  2. Watchers who are teens or young adults: 0.4 + 0.1575 = 0.5575.
  3. Divide by the total probability of watching, 0.58.
  4. 96.1% is above 96%, so the strategy meets the target, but only by a small margin (the 0.15 × 0.15 = 0.0225 adult watchers are about 3.9% of watchers).
  5. Because the margin is so small, small changes in the real proportions could push it below 96%.
Full working
  1. P(T or YAW) = (0.5 × 0.8 + 0.35 × 0.45) ÷ 0.58 | conditional probability
  2. = 0.5575 ÷ 0.58
  3. = 0.9612 ≈ 96.1%
  4. Check: P(AdultW) = 0.0225 ÷ 0.58 = 0.0388 ≈ 3.9%, and 96.1% + 3.9% = 100%
  5. 96.1% ≥ 96%, so yes, the strategy would reach at least 96% of watchers (just)
c(i)
Achieved

Video interaction tree: first video Like 0.35, Comment 0.5, Share the rest; after a share, second video L 0.2, C 0.3; after a second comment, P(Follow) = 0.5. Find P(share first, comment second, then follow).

Answer

0.0225

Step by step
  1. Missing first branch: P(Share first) = 1 − 0.35 − 0.5 = 0.15.
  2. Other missing branches: after a first Like, P(S) = 1 − 0.4 − 0.4 = 0.2; after a first Share, P(S) = 1 − 0.2 − 0.3 = 0.5.
  3. Multiply along Share, then Comment, then Follow.
Full working
  1. P(Share first) = 1 − 0.35 − 0.5 = 0.15branches add to 1
  2. P(S, C, F) = 0.15 × 0.3 × 0.5multiply along the path
  3. = 0.0225
c(ii)
Excellence

After a first comment, three times as many comment on the second video as share it. If 48% of all users comment on the second video, find the probability a random user likes the second video.

Answer

≈ 0.277 (exactly 0.27667, i.e. 83300)

Step by step
  1. Let the probability of sharing the second video after a first comment be s. Then commenting is 3s and liking is 1 − 4s.
  2. Write the overall probability of commenting on the second video using all three first-video branches, and set it equal to 0.48.
  3. Solve for s, which gives the liking probability on the Comment branch.
  4. Then add the three paths that end in a Like on the second video.
Full working
  1. Comment branch: P(S) = s, P(C) = 3s, P(L) = 1 − 4s
  2. P(C₂) = 0.35 × 0.4 + 0.5 × 3s + 0.15 × 0.3 = 0.48total probability of commenting
  3. 0.14 + 1.5s + 0.045 = 0.48
  4. 1.5s = 0.295, so s = 0.19667so P(C) = 0.59, P(L) = 1 − 4(0.19667) = 0.21333 on the Comment branch
  5. P(L₂) = 0.35 × 0.4 + 0.5 × 0.21333 + 0.15 × 0.2add the Like paths
  6. = 0.14 + 0.10667 + 0.03 = 0.27667
  7. Check: P(S₂) = 0.35 × 0.2 + 0.5 × 0.19667 + 0.15 × 0.5 = 0.24333, and 0.27667 + 0.48 + 0.24333 = 1
  8. P(user likes the second video) ≈ 0.277
Question TWO

Question TWO

a(i)
Achieved

Daily teen TikTok time is normal with mean 85 min and standard deviation 20 min. Find P(50 < X < 120).

Answer

0.9199

Step by step
  1. Both 50 and 120 are 35 minutes, or 1.75 standard deviations, from the mean.
  2. Use the normal distribution on a calculator with lower 50, upper 120, μ = 85, σ = 20.
Full working
  1. X ~ N(85, 20²)
  2. z = 50 − 8520 = −1.75, z = 120 − 8520 = 1.75standardise
  3. P(−1.75 < Z < 1.75) = 0.9199
  4. P(50 < X < 120) ≈ 0.920
a(ii)
Achieved

Same model. What percentage spend more than 2 hours a day?

Answer

About 4.0% (4.01%)

Step by step
  1. Change 2 hours into minutes: 120 minutes.
  2. Find the area above 120 under the normal curve.
Full working
  1. 2 hours = 120 min
  2. P(X > 120) = P(Z > 1.75)z = 120 − 8520
  3. = 0.0401
  4. About 4.0% of teenagers
a(iii)
Merit

Same model. Find the greatest daily time for the lowest 15% of teenagers.

Answer

About 64.3 minutes

Step by step
  1. We need the value with 15% of the distribution below it, an inverse normal calculation.
  2. The z value with 0.15 to its left is −1.0364.
Full working
  1. P(X < k) = 0.15
  2. z = −1.0364inverse normal, area 0.15 to the left
  3. k = 85 − 1.0364 × 20k = μ + zσ
  4. = 64.27
  5. The lowest 15% spend at most about 64.3 minutes
b
Merit

Viewing time is normal with standard deviation 15 s, and 90% watch for more than 57 s. Find the mean.

Answer

About 76.2 seconds

Step by step
  1. If 90% are above 57 s, then 10% are below 57 s.
  2. The z value with 0.10 below it is −1.2816.
  3. Use x = μ + zσ and solve for μ.
Full working
  1. P(X > 57) = 0.9, so P(X < 57) = 0.1
  2. z = −1.2816inverse normal, area 0.1 to the left
  3. 57 = μ − 1.2816 × 15
  4. μ = 57 + 19.22 = 76.22
  5. Check: P(X > 57) with N(76.22, 15²) = 0.900
  6. Mean ≈ 76.2 seconds
c
Excellence

Using the histogram of a household's daily TikTok minutes, compare the data with a normal model with σ = 20 min (national figures: mean 95, standard deviation 20). Discuss at least two of centre, shape and spread.

Answer

Shape and spread fit a normal model with σ = 20 well, but the centre is about 69 minutes, well below the national mean of 95, so N(95, 20²) is not a good model for this household; something like N(69, 20²) fits much better.

Step by step
  1. Shape: the histogram is unimodal with its peak in the 60 to 70 minute class and is roughly symmetric, tapering off evenly on both sides. That is the bell shape of a normal distribution.
  2. Centre: from class midpoints the mean is about 69 minutes, and the median is also in the 60 to 70 class. A normal model is centred on its mean, so the household centre is about 26 minutes below the national 95 minutes. The household uses TikTok less than the average New Zealander.
  3. Spread: the estimated standard deviation from the histogram is about 21.7 minutes, close to σ = 20. The data run from 20 to 120 minutes, a range of 100, close to the 6σ = 120 minutes expected to hold almost all normal data.
  4. Check of the 68% rule: about 16 of the 24 values (67%) lie within one standard deviation of the mean (about 49 to 89 minutes), very close to 68%.
  5. Conclusion: a normal model with σ = 20 is reasonable, but it should be centred near 69 minutes, not 95. The sample is small and the days are not independent (the same 4 people each day), so the fit should be treated with caution.
Full working
  1. Class midpoints 25, 35, ..., 115 with frequencies 1, 1, 2, 4, 5, 4, 3, 2, 1, 1, n = 24
  2. Mean ≈ Σfx ÷ n = 1660 ÷ 24 ≈ 69.2 min
  3. Median: 12th and 13th values lie in the 60 to 70 class
  4. s ≈ 21.7 min, close to σ = 20
  5. Within mean ± 20 (about 49 to 89): 4 + 5 + 4 + 3 = 16 values, 1624 ≈ 67%, close to 68%
  6. Shape: unimodal, roughly symmetric, so normal is appropriate; centre about 26 min below 95
  7. Better model: approximately N(69, 20²)
Question THREE

Question THREE

a(i)
Achieved

Using the screentime and sleep table (480 teenagers), what percentage have more than 3 hours of screentime a day?

Answer

25%

Step by step
  1. Use the row total for more than 3 hours and divide by the grand total.
Full working
  1. 120 ÷ 480 = 0.25
  2. 25%
a(ii)
Achieved

Given a teenager has 1 to 3 hours of screentime, find the probability their sleep is not disrupted.

Answer

0.6

Step by step
  1. 'Given' means we only look at the 1 to 3 hours row, which has 240 people.
  2. 144 of them have sleep not disrupted.
Full working
  1. P(not disrupted1 to 3 h) = 144 ÷ 240
  2. = 0.6
a(iii)
Merit

Find the probability a teenager has less than 1 hour of screentime or disrupted sleep.

Answer

0.65

Step by step
  1. Add the less than 1 hour row total and the sleep disrupted column total.
  2. The 30 people in both groups have been counted twice, so subtract them once.
Full working
  1. P(<1 h or D) = (120 + 222 − 30) ÷ 480addition rule
  2. = 312 ÷ 480
  3. Check: 120 + 96 + 96 = 312
  4. = 0.65
b(i)
Merit

Evaluate the claim that the risk of sleep disruption for heavy users (over 3 hours) is more than twice that for all other users, using relative risk.

Answer

Relative risk ≈ 2.29, so for this sample the claim is true: heavy users are about 2.3 times as likely to have disrupted sleep as all other users.

Step by step
  1. Risk for heavy users: 96 out of 120 have disrupted sleep, 0.8.
  2. Risk for all other users: combine the two lower rows, (30 + 96) out of (120 + 240), which is 126360 = 0.35.
  3. Relative risk = 0.8 ÷ 0.35 ≈ 2.29, which is more than 2, so the claim is supported by this data.
  4. Breaking it down: heavy users have 3.2 times the risk of those under 1 hour (0.8 ÷ 0.25), and exactly 2 times the risk of those on 1 to 3 hours (0.8 ÷ 0.4).
  5. So the 'more than twice' only holds against all other users combined; against the 1 to 3 hour group it is exactly twice, not more. The claim is true as worded, but only just, and it depends on which comparison group is used.
  6. This is one sample of 480, so the true relative risk in the population could be a bit above or below 2.29.
Full working
  1. Risk (>3 h) = 96120 = 0.8
  2. Risk (others) = 30 + 96120 + 240 = 126360 = 0.35
  3. RR = 0.8 ÷ 0.35 = 2.286
  4. Compare: 0.8 ÷ 0.25 = 3.2 (vs <1 h); 0.8 ÷ 0.4 = 2.0 (vs 1 to 3 h)
  5. RR ≈ 2.29 > 2, so the claim holds for this sample, but only narrowly
b(ii)
Excellence

Discuss two limitations of the study behind the 'more than twice the risk' claim.

Answer

1. It is an observational survey, so it shows an association, not that screentime causes sleep disruption. 2. Sleep disruption and screentime are self-reported and the sample may not represent all teenagers, so the relative risk may be biased or vary from sample to sample.

Step by step
  1. Causation: teenagers were not randomly given different screentimes. Other factors (for example stress, anxiety, caffeine, or poor sleep making them use phones at night) could explain the link. Poor sleep might even cause more screentime, rather than the other way round.
  2. Self-reporting: 'frequent difficulty falling asleep or waking up' is reported by the teenagers themselves and is subjective. Screentime is also often under or over estimated. Errors in either change the counts and so change the relative risk.
  3. Sample: 480 teenagers from one survey may not represent all teenagers (for example region, age range, or who chose to respond). The relative risk of 2.29 is close to 2, so a different sample could easily give a value below 2.
  4. Categories: 'more than 3 hours' is a wide group that mixes 3.5 hours with 8 hours, and screentime type (gaming, social media, schoolwork) and time of day are not recorded.
  5. Any two of these, explained in context, would be accepted.
Full working
  1. Limitation 1: observational, so association not causation; confounding variables possible
  2. Limitation 2: self-reported, subjective measures, and sample may not be representative; RR ≈ 2.29 is near 2, so sampling variation could change the conclusion

AS 91267 past papers · All NCEA Level 2 Maths

Worked by Vertex Academy teachers. NZQA's marking schedule for 2026 is not out yet; these are our answers, and the grade on each part is our estimate.