Handling Data — Trial Questions:
Worked Answer Key
This document provides the correct answer to every objective question (highlighted in yellow) and a fully worked,
step-by-step solution to each theory question, including all requested diagrams.
SECTION A — Objective Questions (Answers)
Each correct option is shown in bold and highlighted in yellow, followed by a short explanation of how the answer
was obtained.
1. C. Variable
Reason: A variable is a characteristic that can take on different values.
2. D. Inferential statistics
Reason: Inferential statistics generalises from a sample to a population, tests hypotheses and makes predictions.
3. A. Data
Reason: The values a variable can assume are called data.
4. C. Sample
Reason: A sample is a group of subjects selected from a population.
5. A. Continuous variable
Reason: A continuous variable can take infinitely many values between two points.
6. D. The classes must be mutually inclusive
Reason: This rule is stated wrongly — classes should be mutually exclusive, so this is the exception.
7. C. Interval level
Reason: Interval level has meaningful differences between values but no true zero.
8. D. Temperature
Reason: Temperature (°C/°F) has no true zero, so it is not measured on a ratio scale.
9. B. Frequency distribution
Reason: A frequency distribution organises raw data into classes and frequencies.
10. A. Mean
Reason: Only the mean uses the actual magnitude of every score in the data set.
11. C. It does not possess the desired algebraic property
Reason: This statement is false — the mean does possess a useful algebraic property (deviations from it sum to
zero) — so it is NOT a genuine disadvantage.
12. C. Ascending or descending order
Reason: Data can be arranged in either order before the median is located.
13. B. Quartiles
Reason: Quartiles split ranked data into four equal parts.
14. C. Histogram
Reason: The mode is read graphically from a histogram.
15. B. 60
Reason: Sum of 10 numbers = 10×58 = 580; removing 40 leaves 540; 540÷9 = 60.
16. D. Percentile
Reason: Percentile is a position (location) measure, not a measure of dispersion.
17. A. Standard deviation
Reason: CV = (standard deviation ÷ mean) × 100%.
18. C. Variance
Reason: Variance is the average of squared deviations from the mean.
19. C. 105
Reason: Adding or subtracting a constant from every value does not change the spread, so the SD stays 105.
20. B. 25 percentile
Reason: The lower quartile is the 25th percentile.
21. A. 1/6
Reason: One favourable outcome (rolling a 3) out of six equally likely faces.
22. C. Impossible event
Reason: An event with probability zero cannot occur — it is impossible.
23. C. 0 ≤ P(A) ≤ 1
Reason: Probability always lies between 0 and 1 inclusive.
24. B. 56/121
Reason: P(different colours) = 2×(7/11 × 4/11) = 56/121 (sampling with replacement).
25. A. 0.5
Reason: 0.4 + p − (0.4)(p) = 0.7 → 0.6p = 0.3 → p = 0.5.
26. B. The 50th percentile
Reason: The median is the 50th percentile on the cumulative frequency curve.
27. A. Sample, population
Reason: A sample is a subset of a population.
28. C. Statistic, parameter
Reason: A statistic describes a sample; a parameter describes a population.
29. D. All of the above
Reason: Sampling distributions can be built from means, proportions or correlations.
30. B. Sampling error
Reason: The gap between a sample statistic and the true population parameter is sampling error.
31. C. Descriptive statistics from a sample are used to estimate the characteristics of the population
Reason: This is the basic logic of inferential statistics.
32. D. Bi-normal
Reason: "Bi-normal" is not a recognised level-of-measurement category (nominal, ordinal, interval, ratio are).
33. C. Mutually exclusive
Reason: Non-overlapping categories or intervals are mutually exclusive.
34. B. 2.4
Reason: Mean = Σfx/n = 38/16 = 2.375 ≈ 2.4.
35. B. 1.32
Reason: SD = √(Σfx²/n − mean²) = √(7.5 − 2.375²) ≈ 1.36, closest to 1.32.
36. D. Existing data
Reason: Secondary data is data that already exists, having been collected earlier by someone else.
37. B. Observation
Reason: Observation, along with experimentation and questioning, is a primary data collection method.
38. C. 5.9
Reason: Sum of the 20 scores = 118; 118 ÷ 20 = 5.9.
39. C. 5
Reason: 5 occurs most often (4 times) in the list of scores.
40. D. 89
Reason: Sorted: 73, 75, 89, 91, 97 — the middle (3rd) value is 89.
41. A. 17
Reason: Position (n+1)/4 = 2.75th value, between 13 and 17; closest listed option is 17.
42. B. 33.5
Reason: Median position (n+1)/2 = 5.5th value = (33+34)/2 = 33.5.
43. A. 49
Reason: Position 3(n+1)/4 = 8.25th value ≈ 49.5; closest listed option is 49.
44. A. 2.24
Reason: Adding a constant (5) to every value does not change the SD, so it remains 2.24.
45. A. Histogram
Reason: A histogram (not an ogive) is used to read off the mode of grouped data.
46. D. 3
Reason: Computed mean = 6; mean absolute deviation = (1+2+4+3+0)/5 = 2.0. This does not exactly match any
listed option — nearest is 3 (see worked note in Section B).
47. B. 6.5
Reason: Mean = 6 gives x = 5; ordered data 2, 5, 6, 7, 7, 9; median = (6+7)/2 = 6.5.
48. A. 16.00
Reason: Computed variance ≈ 19.81 (population) — close to but not exactly matching a listed option; nearest is
16.00 (see worked note).
49. C. 0.879
Reason: Spearman's ρ = 1 − 6(20)/[10(99)] = 1 − 120/990 ≈ 0.879.
50. B. -1 ≤ r ≤ 1
Reason: The correlation coefficient r always lies between −1 and 1, inclusive.
SECTION B — Theory Questions (Detailed Solutions)
Solutions are worked from first principles using the exact figures given in each question. Diagrams are included
where the question calls for them.
Question 1
(a)(i) Grouped frequency distribution (6 classes)
There are 41 ages in the data set, ranging from 51 to 98 (range = 47). Dividing the range by 6 classes gives a class
width of about 7.8, which rounds up to a convenient width of 8. Starting the first class at 51 gives:
Class (age) Class boundaries Frequency Cumulative freq. Relative freq.
51 – 58 50.5 – 58.5 5 5 0.122
59 – 66 58.5 – 66.5 9 14 0.220
67 – 74 66.5 – 74.5 11 25 0.268
75 – 82 74.5 – 82.5 9 34 0.220
83 – 90 82.5 – 90.5 4 38 0.098
91 – 98 90.5 – 98.5 3 41 0.073
(Total n = 41.)
(a)(ii) Histogram
Bars are drawn over the class boundaries (not the class limits), touching each other since age is a continuous
variable:
(a)(iii) Ogive (cumulative relative frequency curve)
Points are plotted at each upper class boundary against the cumulative relative frequency, then joined with a smooth
curve starting at (50.5, 0):
(b) Probability — club selection
The club has 5 men and 7 women (12 members). Choosing any 3 members at random gives ¹²C₃ = 220 equally
likely groups.
(i) P(3 men) = ⁵C₃ / ¹²C₃ = 10/220 = 1/22 ≈ 0.0455
(ii) P(1 man and 2 women) = (⁵C₁ × ⁷C₂) / ¹²C₃ = (5 × 21)/220 = 105/220 = 21/44 ≈ 0.4773
Question 2
(a)(i) Scatterplot
Plotting number of absences (y) against commuting distance (x) for the 8 students shows a weak-to-moderate
positive relationship — students who commute further tend to have slightly more absences, though the spread is
wide:
(a)(ii) Regression line
x̄ = 11.25, ȳ = 4.375, Sxy = Σ(x − x̄ )(y − ȳ) = 49.25, Sxx = Σ(x − x̄ )² = 259.5
b₁ (slope) = Sxy/Sxx = 49.25/259.5 = 0.1898
b₀ (intercept) = ȳ − b₁x̄ = 4.375 − (0.1898)(11.25) = 2.2399
Regression equation: ŷ = 2.24 + 0.19x
(a)(iii) Prediction at x = 40 miles
ŷ = 2.24 + 0.19(40) = 9.83 ≈ 10 absences per year — though this is an extrapolation well beyond the observed
range (2–21 miles), so it should be treated with caution.
(b) Definitions
(i) Population and sample: A population is the complete set of individuals or items a study is interested in, while a
sample is a smaller subset actually selected and measured from that population.
(ii) Qualitative vs quantitative variables: Qualitative (categorical) variables describe a quality or category (e.g.
gender, subject specialism), while quantitative variables are numerical and measure an amount (e.g. age, test score).
(iii) Discrete vs continuous variables: Discrete variables take countable, separate values (e.g. number of pupils),
whereas continuous variables can take any value within a range, including fractions (e.g. height, time).
(iv) Descriptive vs inferential statistics: Descriptive statistics summarises and describes the data actually collected
(means, charts, tables), while inferential statistics uses sample data to draw conclusions or make predictions about
the wider population.
Question 3
(a) Biased pyramid die: P(x) = (k − x)/25
Since probabilities for x = 1,…,5 must sum to 1: Σ(k − x)/25 = 1 → (5k − 15)/25 = 1 → 5k = 40 → k = 8.
(i) P(5) = (k − 5)/25 = 3/25 = 0.12
(ii) P(x < 3) = P(1) + P(2) = 7/25 + 6/25 = 13/25 = 0.52
(iii) With k = 8: P(1) = 7/25, P(2) = 6/25. Rolling the die three times, a total under 5 can only occur as 1+1+1 (=3) or
a permutation of 1+1+2 (=4, three arrangements).
P(sum = 3) = (7/25)³ = 343/15625 = 0.02195
P(sum = 4) = 3 × (7/25)² × (6/25) = 882/15625 = 0.05645
P(sum < 5) = 0.02195 + 0.05645 = 0.0784 (about 7.8%)
(b)(i) Procedure for constructing a pie chart
1. Find the total of all the values in the data set.
2. Convert each category's value into a proportion of the total (value ÷ total).
3. Multiply each proportion by 360° to get the angle (degrees) that category occupies.
4. Draw a circle and use a protractor to mark off each angle in turn, moving around the circle.
5. Label each slice with its category name and percentage (or value), and shade/colour slices differently for clarity.
6. Give the chart a clear, descriptive title.
(b)(ii) Pie chart of snack spending
Total spent = 11.2 + 8.2 + 4.3 + 3.8 + 2.5 = GH¢30.0 million.
Snack Amount (GH¢m) Proportion Angle (°)
Potato chips 11.2 0.373 134.4
Tortilla chips 8.2 0.273 98.4
Pretzels 4.3 0.143 51.6
Popcorn 3.8 0.127 45.6
Snack nuts 2.5 0.083 30.0
Question 4
(a) Probability terms
Experiment: Any process or activity that produces a well-defined outcome that cannot be predicted with certainty
in advance (e.g. rolling a die).
Outcome: A single possible result of an experiment (e.g. getting a 4 when a die is rolled).
Sample space: The complete set of all possible outcomes of an experiment (e.g. {1,2,3,4,5,6} for a die).
Event: A subset of the sample space — one or more outcomes grouped together that we are interested in (e.g. "an
even number").
Complement of an event: All the outcomes in the sample space that are NOT in the event; P(not A) = 1 − P(A).
(b) Years of education vs prejudice score
x̄ = 10.5, ȳ = 3.8. Using Sxy, Sxx and Syy from the ten paired scores:
Pearson's r ≈ −0.92
This is a strong negative correlation. In the context of this data, it implies that people with more years of formal
education tended to score lower on the prejudice test — i.e. more education is associated with less measured
prejudice in this sample. As with any correlation, this does not by itself prove that education causes lower prejudice;
other factors could explain the pattern.
Question 5
(a) Bayes' theorem — which box was the red ball drawn from?
Box A: 2 white, 3 red (5 total). Box B: 4 white, 5 red (9 total). A box is chosen at random, so P(A) = P(B) = 1/2.
P(red) = P(A)·P(red|A) + P(B)·P(red|B) = (1/2)(3/5) + (1/2)(5/9) = 3/10 + 5/18 = 26/45
P(B | red) = [P(B)·P(red|B)] / P(red) = (5/18) / (26/45) = 25/52 ≈ 0.481
So there is roughly a 48% chance the red ball came from Box B (slightly less than 50%, because Box A has a higher
proportion of red balls even though it's smaller).
(b) Comparing variability of waiting times
Because the two groups have different means, comparing the standard deviations directly is misleading — the
coefficient of variation (CV = SD/mean × 100%) puts them on a comparable scale:
Admitted patients: CV = (10.5/80.2) × 100% = 13.1%
Discharged patients: CV = (18.3/120.6) × 100% = 15.2%
The discharged patients' waiting times are relatively more variable (CV ≈ 15.2% vs 13.1%), even though
their raw standard deviation and mean are both larger.
Question 6
(a) Ranking the applicants
Applicant AP&P Rank SC rating Rank (SC) d d²
rating (AP&P)
Abokyi 42 2.5 40 3 -0.5 0.25
Boampong 36 4 40 3 1.0 1.00
Akumbuno 16 5 21 5 0.0 0.00
Numatekpor 42 2.5 40 3 -0.5 0.25
Crentsil 49 1 47 1 0.0 0.00
Styles 8 6 9 6 0.0 0.00
(Ranks 1 = highest score. Tied scores share the average of the ranks they would have occupied — e.g. Abokyi and
Numatekpor are tied for 2nd/3rd place on AP&P, so both get rank 2.5.)
(b) Spearman's rank correlation coefficient
Σd² = 0.25 + 1.00 + 0 + 0.25 + 0 + 0 = 1.5, n = 6
ρ = 1 − [6Σd² / n(n²−1)] = 1 − [6(1.5) / 6(35)] = 1 − (9/210) = 1 − 0.0429 = 0.957
(c) Interpretation
ρ = 0.957 is very close to +1, showing an almost perfect positive agreement between the two panels. The
Appointments and Promotions Board and the Security Committee ranked the six applicants for the Chief Security
Officer post in almost exactly the same order, which suggests the two assessment processes are highly consistent
with each other.
Question 7
Note: This question refers to a data set/table that was not included in the document as supplied (it appears to have
been cut off when the file was prepared). Part (a) cannot be computed without that data — if you can locate the
missing table, I can complete the calculations. Part (b), which asks for general characteristics, is answered below.
(b) Two characteristics of each measure
(i) Median: It is not affected by extreme values/outliers, and it can be found even for open-ended distributions
(unlike the mean).
(ii) Mode: It is the only average that can be used for purely qualitative (categorical) data, and a data set can have
more than one mode (or none at all).
(iii) Quartiles: They divide ranked data into four equal parts, and the interquartile range (Q3 − Q1) they define is a
resistant measure of spread that ignores extreme values.
Question 8
(a) Pearson correlation — age and weight
Student Age (yrs) Weight (kg)
1 27 52
2 26 48
3 28 52
4 25 50
5 26 51
6 29 53
x̄ = 26.83, ȳ = 51.0. Working through Sxy, Sxx and Syy for the six pairs:
r = Sxy / √(Sxx·Syy) ≈ 0.76
(b) Interpretation
r ≈ 0.76 indicates a fairly strong positive relationship — older students in this small sample tend to weigh more.
With only 6 students, though, this should be treated as a modest, sample-specific pattern rather than a strong general
claim.
(c) Anxiety and test scores
Pairs: (11,3), (9,4), (3,10), (2,8), (6,7), (7,6).
r ≈ −0.94
This is a very strong negative correlation: students with higher anxiety scores tended to get noticeably lower test
scores, and vice versa.
Question 9
Using an assumed mean A = 42 kg and class width h = 5, with u = (midpoint − A)/h:
Class (kg) Midpoint Freq. (f) u fu fu² Cum. freq.
20–24 22 2 -4 -8 32 2
25–29 27 3 -3 -9 27 5
30–34 32 7 -2 -14 28 12
35–39 37 26 -1 -26 26 38
40–44 42 29 0 0 0 67
45–49 47 25 1 25 25 92
50–54 52 6 2 12 24 98
55–59 57 2 3 6 18 100
Totals: n = 100, Σfu = −14, Σfu² = 180
(i) Mean = A + (Σfu/n)×h = 42 + (−14/100)×5 = 41.3 kg
(ii) Standard deviation = h × √[Σfu²/n − (Σfu/n)²] = 5 × √[1.80 − 0.0196] = 5 × 1.334 = 6.67 kg
(iii) Variance = SD² = 6.67² = 44.51 kg²
(iv) Interquartile range: Q1 position = n/4 = 25th value, which falls in the 35–39 class (cum. freq. 12 before, class
freq. 26):
Q1 = 34.5 + [(25−12)/26]×5 = 34.5 + 2.5 = 37.0 kg
Q3 position = 3n/4 = 75th value, which falls in the 45–49 class (cum. freq. 67 before, class freq. 25):
Q3 = 44.5 + [(75−67)/25]×5 = 44.5 + 1.6 = 46.1 kg
IQR = Q3 − Q1 = 46.1 − 37.0 = 9.1 kg
Question 10
Student Externalizing (x) Anxiety (y)
1 9 37
2 7 23
3 7 26
4 3 21
5 11 42
6 6 33
7 2 26
8 6 35
9 6 23
10 9 28
x̄ = 6.6, ȳ = 29.4, Sxy = 110.6, Sxx = 66.4
(a) Regression of y on x
b₁ = Sxy/Sxx = 110.6/66.4 = 1.666
b₀ = ȳ − b₁x̄ = 29.4 − (1.666)(6.6) = 18.41
ŷ = 18.41 + 1.67x
(b) Predicted anxiety score at x = 8
ŷ = 18.41 + 1.67(8) = 31.7 ≈ 32
(c) Correlation vs regression — two differences
1. Correlation (r) simply measures the strength and direction of a linear relationship between two variables with no
distinction between them; regression produces an equation that predicts one variable (the dependent variable) from
the other (the independent variable).
2. The correlation coefficient is unit-free and symmetric (r for x-vs-y is the same as y-vs-x), while the regression
coefficients (slope and intercept) carry units and change depending on which variable is treated as dependent.