Sixth Term Examination Paper [STEP]

Mathematics 2 [9470]

2020

Examiner’s Report

Worked Solutions

©UCLES 2020
This document was initially designed for print and as such does not reach
accessibility standard WCAG 2.1 in a number of ways including missing text
alternatives and missing document structure.

If you need this document in a different format please email

[email protected] telling us your name, email
address and requirements and we will respond within 15 working days.

©UCLES 2020
 STEP MATHEMATICS 2
2020
Examiner’s Report

©UCLES 2020
 STEP 2: Examiner's Report

There were just over 800 entries for this paper, and good solutions were seen to all of the
questions. Candidates should be aware of the need to provide clear explanations of their
reasoning throughout the paper (and particularly in questions where the result to be shown
is given in the question). Short explanatory comments at key points in solutions can be very
helpful in this regard, as can clearly drawn diagrams of the situation described in the
question. The paper included a few questions where a statement of the form “A if and only
if B” needed to be proven – candidates should be aware of the meaning of such statements
and make sure that both directions of the implication are covered clearly.
In general, candidates who performed better on the questions in this paper recognised the
relationships between the different parts of each question and were able to adapt methods
used in earlier parts when working on the later sections of the question.

©UCLES 2020
Question 1
This was the most popular question on the paper, and also the one on which candidates
performed the best.
In general, candidates were confident in applying the substitution given in part (i) and
completed the integration correctly, although there were a number of solutions in which
careless errors were seen.
Successful completion of the remaining two parts required candidates to understand the
reason why the substitution suggested in part (i) was helpful, and so candidates who
continued to apply the same substitution from part (i) to later parts were often unable to
make useful progress, particularly on part (iii).
In part (ii) many candidates took the approach of making a sequence of two substitutions to
reach the answer. Attempts involving just one substitution were more likely to include
errors, although a number of these were also completed successfully. In some cases
candidates recognised that the integral was going to be reduced to a form similar to that in
part (i), but then did not obtain exactly the correct function to be integrated.
In part (iii) candidates who were continuing with the same substitution as in part (i) often
spent a lot of time rearranging the function to be integrated without success. Some
attempts to apply partial fractions were seen in this section despite the fact that the factors
were square roots of linear expressions. Many of those who were able to adapt the
substitution from part (i) did recognise the form of the new function to be integrated and
often selected an appropriate substitution.

©UCLES 2020
Question 2
This was a popular question and candidates in general achieved good marks. Most
candidates approached part (i) with reasonable confidence and followed the question’s
intended path of separating variables and integrating directly. Of these, the majority failed
1
to consider the integral of as ln|x|. This was entirely understandable for work on C1 as the
x
question specifies that it lies entirely in the first quadrant. While ln⁡(xy) was okay to deal
with in the case where both x and y were negative, it would be good for candidates to be
clear about the way in which the modulus function is being dealt with here.
Around 1 in 5 candidates took the alternate route of differentiating the given answer to
show that it fitted the differential equation in the question. Unfortunately, not only is this
much more demanding work, but almost all such attempts failed to realise the need to
check the initial condition x = y = 1 as part of the solution. Candidates should be clear about
the distinction between “show” and “verify” in such questions.
In part (ii), the sketching of two fairly straightforward functions caused unexpected
difficulties when it came to putting them together suitably on the same diagram; it was
important to show that the two curves intersect twice and many sketches failed to have
them crossing more than once, if at all. This made the subsequent reasoning and sketch
of C1 very awkward. Even for those who got this far entirely successfully, it was a common
problem to find the sketch of C1 drawn without the helpful guidance supplied within the
question and any results arising from correct working to date. In particular, it was important
for candidates to demonstrate the symmetry in the line y = x and the restrictions provided
by the lines x + y = 2 and x + y = 4, all of which really should have appeared on the diagram.
Many attempts petered out by the time of part (iii), and very few candidates attempted to
consider a graphical method (directly analogous to the method promoted in part (ii)) to
show that the curve of C2 was constrained by the line x + y = –2. By the time they came to
draw this second solution to the original differential equation, many candidates had
forgotten either or both of the given bits of information; namely, that there was symmetry
in y = x and that C2 existed only in the 3rd-quadrant. Many candidates just assumed
that C2 was the reflection of C1.
Candidates should be advised not to attempt to sketch curves by plotting points, as was
seen in a number of cases. Instead, the information established in earlier parts of the
question should be used to ensure that the key points are marked and that the shape is
correct.

©UCLES 2020
Question 3
This was the second least attempted of the pure questions. Relatively few candidates made
a complete attempt at all of the parts and only 4 achieved full marks for the question.
The question consisted of a succession of given results which were to be established. Thus,
candidates needed to be more aware of the importance of providing careful and thorough
explanations and justifications for each step that they took along the way. Many marks were
lost as a result of carelessness in providing all of the necessary details.
A significant number of candidates thought that the implication in (i) showed that the
sequence was either increasing or decreasing and so got little or no credit. Establishing the
given relations in (i) was generally done quite well, with candidates demonstrating a
considerable range of algebraic skills in their working. But then a lot of candidates failed to
show that the sequence was positive, which undermined their attempts to deduce that the
sequence was unimodal.
Many candidates used an induction proof for the first proof in part (ii) despite the fact that
a more direct approach was possible and considerably simpler. The “asked-for” induction
proof was usually handled well, though establishing the baseline case was often flawed;
many either overlooked the need to establish both of the cases n = 1 and n = 2 or, when
giving a one-step induction proof with the help of the previously-established result, chose
an incorrect baseline case.
The final part of the question was often avoided, though full attempts often gained full
credit. Again, the usual oversight was to fail to establish positivity. Many of those who
produced only a faltering solution here overlooked the need to compare successive terms,
usually merely working with an expression for ur only, often with the use of differentiation -
attempts along such lines invariably lost all of the final 7 marks allocated, primarily because
the required result is based on discrete values of r while calculus works with continuous
values.

©UCLES 2020
Question 4
Most candidates could justify the triangle inequality in part (i) (as well as arguing that the
shortest distance is the straight line, there were successful uses of the cosine rule or
𝑐 = 𝑎 cos 𝐵 + 𝑏 cos 𝐴), but were less confident in proving the converse in part (ii).
Successful approaches were to consider two circular loci for the SSS construction, or to fix
two sides and vary the angle between them; in both cases care was needed to ensure all
three inequalities were actually used, for example checking that neither circle can contain
the other, which was often omitted. There was one elegant solution using three pairwise
tangent circles.
A reasonable number of candidates obtained correct answers of “always”, “sometimes” (by
examples) and “never” for part (iii) A, B and C respectively, although it was surprisingly
common to forget that 𝑎, 𝑏, 𝑐 must be the sides of a triangle. However, B caused some
confusion as many candidates spent some time trying to prove that the new lengths did
satisfy the triangle inequality.
Parts (iii) D and (iv) were found much harder and many candidates did not attempt them. A
reasonable number of candidates were able to make some progress, but there were few
complete solutions to either of these parts. Common errors for (iii) D were showing that the
sum of the three inequalities is a true statement, or attempting to prove a positive result by
examples. Most substantial attempts at (iv) used a different approach of fixing the order of
𝑎, 𝑏, 𝑐 and reducing the problem to proving one of the three inequalities.

©UCLES 2020
Question 5
This was a popular question and many of the solutions made good progress on the early
parts of the question.
The majority of candidates gained full marks for part (i), but some candidates did not
mention that 𝑥 − 𝑑(𝑥) ≥ 0.
There was a wide range of marks achieved on part (ii). The proof that 𝑥 − 44𝑑(𝑥) is a
multiple of 9 if and only if 𝑥 is a multiple of 9 was completed well by those who managed to
prove the result, but the majority of other attempts seen did not score any marks. In a small
number of cases only the “if” direction was proved. Those who were unable to prove the
first result in this part were often able to continue and find the required bounds on 𝑥
however. Candidates who had completed both of these parts generally managed to find the
correct answer 𝑥 = 792, but did not necessarily fully justify that it was the only one.
Most candidates scored low marks on part (iii). It was very common to see an insufficient
proof that 9|𝑥. Without guidance from the question as to how to find bounds on 𝑥, students
produced a wide range of approaches; better bounds were needed if the student only used
107|𝑥, but the simple bound 𝑑(𝑑(𝑥)) ≤ 𝑑(𝑥) together with divisibility by 963 was
sufficient.

©UCLES 2020
Question 6
This was the second most popular question on the paper. In general, candidates need to be
careful when proving statements of the form “A if and only if B” and should be aware that in
some cases it may not be possible to prove both directions in one go. Candidates should
also be aware that, in some cases, the algebra is not sufficient on its own to demonstrate
the reasoning and explanations of the steps are often helpful.
Part (i) of the question was generally completed well. In part (ii) many largely successful
attempts were seen, but few candidates picked up all of the marks for this section. The main
errors arose from not adequately considering cases and so dividing by 0, and from not
noticing that 𝑎2 = 𝑑2 = 1 could result in 𝑎 = −𝑑 = ±1. The most successful attempts in
this part were the ones that separated the two directions of the implications. Many
candidates misused the condition 𝑀 ≠ ±𝐼 in trying to prove the implication in one direction
or did not check this condition when proving the implication in the opposite direction.
Few attempts at part (iii) were seen and a common mistake was to do the component-wise
algebra to find 𝑀4 instead of using the results from previous parts. In general, those who
had understood the previous parts and attempted part (iv) were able to solve the final part
of the question.

©UCLES 2020
Question 7
This was the least popular of the pure questions and also the one on which marks were
lowest on average.
Many candidates were able to show the first result, that |𝑤 − 1| is independent of 𝑡.
However, candidates often did not explain well enough the connection between the form of
𝑧 and the line 𝑅𝑒(𝑧) = 3.
The next part of part (i) required noting that the centre lies on the real axis and working out
|𝑤 − 𝑐|. Some candidates guessed the value of 𝑐. Common mistakes here included guessing
𝑐 = 1, 𝑝 − 2, or failing to note conditions in which |𝑤 − 𝑐| is independent of 𝑡. In many
solutions the absolute value sign on the radius was forgotten.
Part (ii) was similar to the previous part but required noting that the centre lies on the
imaginary axis and working out |𝑤 − 𝑐𝑖|. In both parts a common attempt was the guess the
centre to be at a point 𝑧 = 𝑎 + 𝑏𝑖, few candidates were successful using this method. Again,
absolute value signs on the radius were regularly forgotten.
Another successful method employed by candidates in all parts of the question was to use
𝜃 𝜃 𝜃
the substitution 𝑡 = tan , t⁡= (𝑝 − 2) tan , 𝑡 = 𝑞 tan + 2 and using various trig
2 2 2
identities to achieve the centre and radius. A few students also expressed t in terms of
Re(w) and Im(w) and used that to obtain the equation of a circle in ℝ2 .

©UCLES 2020
Question 8
For the first part candidates were asked to sketch a curve 𝑦 = 𝐹(𝑥) based on some
information about the function 𝑓(𝑥). A not insignificant number of candidates instead
sketched 𝑦 = 𝑓(𝑥) but those who sketched the correct curve often earned most of the
marks. When justifying the given form of 𝐹(𝑥) some good explanations were provided, but
in many cases the repeated roots at 𝑥 = 0 and 𝑥 = 𝑐 were not explained. The final section
of part (i) was generally completed well by those who reached it.
The next part was found to be more difficult with many candidates mistakenly using the
local maximum of 𝐹(𝑥) at 𝑥 = 𝑏 to justify the first inequality instead of the local minimum
at 𝑥 = 𝑎. It was common to see justification such as |𝐹(𝑥)|<𝐹(𝑏) without showing first that
𝐹(𝑏) = −𝐹(𝑎). Candidates who spotted the connection with part (i) and substituted 𝑥 = 𝑏
into their expression for 𝐹(𝑥) + 𝐹(𝑐 − 𝑥) were usually able to show 𝑐 − 𝑏 = 𝑎 or 𝑐 > 2ℎ.
For the last section of part (ii), those who realised the connection with the first paragraph
had no issues.
Candidates who reached the final part of the question were often able to obtain expression
for 𝑓(𝑥) and most realised that they needed to calculate 𝑓′′(𝑥) in order to find the
inflection points. However, the final mark for spotting that the roots of 𝑓 ′′ (𝑥) = 0 are
necessarily roots of 𝑓(𝑥) = 0 without explicitly calculating them (and thereby wasting time)
eluded the majority of candidates who reached this part.

©UCLES 2020
Question 9
This was a question that was found to be difficult.
In general, this question was not attempted well, with very few candidates progressing past
the first section. Most candidates managed to pick up all the marks in the initial section of
the question. However, a significant minority of students could not set up the
problem correctly or knew a lot of linear acceleration (suvat) equations but could not apply
them correctly (for example mistaking displacement for position) and received zero
marks. Some candidates eliminated t in favour of x and could not progress to the
last calculation.
Around half the candidates picked up full marks for part (i). However, many candidates tried
to reason with words – almost always unsuccessfully, often believing that the particle
projected from point A could not pass through the line AB.
Most of the candidates received zero marks for part (ii), failing to realise that the result
follows from the height of the particle at the time of collision being non-
negative. Some tried to use conservation of momentum or energy, or the equation
v 2 = u2 + 2as due to the answer being suggestive of velocity squared. Candidates who
were able to progress well on this part generally achieved all of the marks.
Very few candidates progressed to part (iii) and the attempts were often poor. Candidates
who did know how to proceed to the result often did not justify bounds they used to obtain
inequalities.
A significant number of candidates attempted the final part of the question having omitted
earlier parts. In many cases candidates did not fully appreciate the requirements when
asked to show a statement of the form “A if and only if B”.

©UCLES 2020
Question 10
This was the least popular question on the paper and was found to be quite difficult by
those who did attempt it.
It is very useful in questions of this type to produce a good sketch of the situation described
and, unfortunately, many attempts did not do this. The candidates who made good progress
did tend to have good diagrams.
When resolving forces, it is useful to consider the different directions that could be chosen.
In the case of this question, many candidates chose to resolve horizontally and vertically and
as a result produced more complicated equations to deal with. While in most cases they
were able to simplify the equations that they obtained, this method did result in more work
than was necessary. When dividing equations and inequalities by expressions, candidates
should be careful to consider whether the expressions are known to be non-zero (and in the
case of inequalities that it is known whether it is positive or negative).
In the second part of the question, candidates struggled with the more complicated
algebraic expressions that had to be manipulated and many candidates gave up before
reaching the end of the question.

©UCLES 2020
Question 11
This was the more popular of the probability questions and many good attempts were seen,
although the majority were incomplete and only attempted the first one or two parts. Some
candidates made errors when dealing with the conditional probabilities, often thinking for
example that 𝑃(𝐴⁡𝑤𝑖𝑛𝑠) could be obtained by adding 𝑃(𝐴⁡𝑤𝑖𝑛𝑠|𝐻⁡𝑓𝑖𝑟𝑠𝑡) and
𝑃(𝐴⁡𝑤𝑖𝑛𝑠|𝑇⁡𝑓𝑖𝑟𝑠𝑡). In general, those that were able to work confidently with the
conditional probabilities were able to perform very well on this question.
In the first part, a number of candidates failed to consider that games could begin with
either heads or tails when showing that the probability that the game never ends is 0.
1
Additionally, some candidates assumed that 𝑝 = 𝑞 = for the first part of the question,
2
although they often did then use correct expressions in terms of 𝑝 and 𝑞 in the later parts of
the question.
In part (ii) some candidates tried to find a way to enumerate all possible sequences for any
total number of flips, but this approach almost always resulted in some of the possible cases
being omitted.
Many candidates failed to spot that the solution to the third part could be found by an
analogous method to that used in part (ii) and so in many cases no attempt was made at this
final part.

©UCLES 2020
Question 12
This question had the second smallest number of attempts on the paper. Many of these
successfully completed the first part of the question, but then made little progress in the
later sections.
Part (i) was generally well done, although some candidates did not appreciate that ∑ 𝜀𝑖 = 0
when determining whether a fair or biased die is more likely to show the same score on two
successive rolls. Where candidates were able to see the connection between the first and
second parts of the question, the required result was generally proven clearly.
Part (iii) could be approached in a similar way to part (i), but many of the candidates who
reached this point failed to deal with the more complicated terms that arise from the
expansion. Correct solutions to this part were generally very well set out.

©UCLES 2020
 STEP MATHEMATICS 2
2020
Worked Solutions

©UCLES 2020
 STEP 2: BRIEF SOLUTIONS

Only penalise missing +c once in parts (i) and (ii)

1(i)   1  u 2
1 1
 dx   du
 32  1 2
 u 2 1  u 
1
 x  x  1 2

𝑑𝑢 𝑑𝑥
Must include attempt at (or )
𝑑𝑥 𝑑𝑢

1
 2u 2
1
 x  1 2
 2  c
 x 

1(ii) Let 𝑥 − 2 = 𝑠

 1  1
Then 
 dx   3 ds
3 1  1
  x  2  2  x  1 2  s 2 s  3 2

3
Let s =
u 1

 1   u  12 3 1
 d s   du   32 u 2
 23 1  2 2  u  1
1 2
 s s  3 2  3 u
1 1
 3 2
2s  x  1 2
 3    32   c
 s   x 2
1(iii) 1 u
Let x =
u
Allow substitution leading to two algebraic factors in the denominator.

 0
 1  u2  1 
 dx     2  du
1 1  1 1
u 
2  x  1 x  2  2  3 x  2  2 1 1  u  2  3  u  2

If done through a sequence of substitutions:

a further substitution leading to a square root of a quadratic as the
denominator.

1
 1
 du
 1


0 3  2u  u 2  2

©UCLES 2020
 1

1
 du
 
1
 2
0 4  1  u 
2

1
  1  u 
 arcsin  
  2 0

 21   61   31 

©UCLES 2020
 2(i) 1  ky dy kx  1

y dx x

ln y  ky  kx  ln x  c

Hence, ln xy  k  x  y   c

xy  41  x  y    x  y    Ae  
2 2 k x y
 
2 2
C1 is  x  y    x  y   2x  y

2 2
C2 is  x  y    x  y   2x  y  4

In both cases, the equation is invariant under (𝑥, 𝑦) ↦ (𝑦, 𝑥), so symmetrical in
𝑦 = 𝑥.

2(ii)

(4,16)

(2,4)

Graphs: Correct shapes of curves

Graphs: Intersections at (2,4) and (4,16)
(𝑥 − 𝑦)2 ≥ 0, so (𝑥 + 𝑦)2 > 2𝑥+𝑦

Therefore, (𝑥 + 𝑦) must lie between 2 and 4

Graph: Symmetry about 𝑦 = 𝑥

Graph: Closed curve lying between 𝑥 + 𝑦 = 3 ± 1
Graph: Passes through (1,1) and (2,2)

©UCLES 2020
 2(iii) Sketches of 𝑦 = 𝑥 2 and 𝑦 = 2𝑥+4
𝑥 2 > 2𝑥+4 only when 𝑥 < −2.

v
Graph: Symmetry about 𝑦 = 𝑥
Graph: Passes through (-1,-1)
Graph: 𝑦 → 0 as 𝑥 → ∞, 𝑦 → −∞ as 𝑥 → 0

©UCLES 2020
 3(i) Suppose, ∃𝑘: 2 ≤ 𝑘 ≤ 𝑛 − 1 such that 𝑢𝑘−1 ≥ 𝑢𝑘 , but 𝑢𝑘 < 𝑢𝑘+1

Since all of the terms are positive, these imply that 𝑢𝑘2 < 𝑢𝑘−1 𝑢𝑘+1 , so the sequence
does not have property L.
Therefore, if the sequence has property L, once a value 𝑘 has been reached such that
𝑢𝑘−1 ≥ 𝑢𝑘 , it must be the case that all subsequent terms also have that property (which
is the given definition of unimodality).
3(ii) ur   ur 1    ur 1   u r 2  , so ur   ur 1   r 2 u2   u 1 


ur2  ur 1 u r 1  ur2  ur 1 2 ur   2 u r 1  2
  ur   ur 1  for r ⩾ 2

The first identity shows that 𝑢𝑟 > 0 for all r if 𝑢2 > 𝑎𝑢1 > 0.
Since the right hand side of the second identity is always non-negative, the sequence has
property L, and is hence unimodal.
3(iii) 𝑢1 = (2 − 1)𝛼 1−1 + 2(1 − 1)𝛼 1−2 = 1, which is correct.
𝑢2 = (2 − 2)𝛼 2−1 + 2(2 − 1)𝛼 2−2 = 2, which is correct.
Suppose that:
𝑢𝑘−2 = (4 − 𝑘)𝛼 𝑘−3 + 2(𝑘 − 3)𝛼 𝑘−4 , and
𝑢𝑘−1 = (3 − 𝑘)𝛼 𝑘−2 + 2(𝑘 − 2)𝛼 𝑘−3 .

𝑢𝑘 = 2𝛼 ((3 − 𝑘)𝛼 𝑘−2 + 2(𝑘 − 2)𝛼 𝑘−3 ) − 𝛼 2 ((4 − 𝑘)𝛼 𝑘−3 + 2(𝑘 − 3)𝛼 𝑘−4 )
= 𝛼 𝑘−1 (6 − 2𝑘 − 4 + 𝑘) + 𝛼 𝑘−2 (4𝑘 − 8 − 2𝑘 + 6)
= 𝛼 𝑘−1 (2 − 𝑘) + 2𝛼 𝑘−2 (𝑘 − 1)
which is the correct expression for 𝑢𝑘
Hence, by induction ur = (2 − r)αr−1 + 2(r − 1)αr−2

  
ur  u r 1   2  r   r 1  2  r  1  r 2  1  r   r  2r  r 1 
  r 2  2  r  1   2  3r    (r  1)  2

 2N  r  1   2  3r  N  N  1  (r  1)  N  12 
 r 2 2
 2
N
 r 2

N2
(r  1)  rN  N  2

 r 2  N  1
when r = N, uN  u N 1  0
N2

2 r 2
when r = N – 1, uN 1  u N  0
N2
so ur is largest when r = N

©UCLES 2020
 4(i) The straight line distance between two points must be less than the length of any other
rectilinear path between the points.
4(ii)

𝑎
𝑐

Diagram showing two circles and straight line joining their centres.
Length of line and radii of circles are 𝑎, 𝑏 and 𝑐 in some order.
Either statement that the straight line is the longest of the lengths, or explanation that one
circle cannot be contained inside the other.
Explanation that the circles must meet.
4(iii) (A)
If 𝑎 + 𝑏 > 𝑐 then (𝑎 + 1) + (𝑏 + 1) > 𝑐 + 2 > 𝑐 + 1 et cycl., so 𝑎 + 1, 𝑏 + 1, 𝑐 + 1 can
always form the sides of a triangle.
(B)
If 𝑎 = 𝑏 = 𝑐 = 1 we have 1, 1, 1 which can form the sides of a triangle.
1
If 𝑎 = 1, 𝑏 = 𝑐 = 2 we have , 1, 2 which cannot form the sides of a triangle.
2
𝑎 𝑏 𝑐
Therefore, , , can sometimes, but not always form the sides of a triangle.
𝑏 𝑐 𝑎

(C)
If 𝑝 ≥ 𝑞 ≥ 𝑟 then |𝑝 − 𝑞| + |𝑞 − 𝑟| = 𝑝 − 𝑞 + 𝑞 − 𝑟 = 𝑝 − 𝑞 = |𝑝 − 𝑟|
So two of |𝑝 − 𝑞|, |𝑞 − 𝑟|, |𝑝 − 𝑟| will always sum to the third, so they never form the
sides of a triangle.
(D)
If 𝑎 + 𝑏 > 𝑐 then 𝑎2 + 𝑏𝑐 + 𝑏 2 + 𝑐𝑎 = 𝑎2 + 𝑏 2 − 2𝑎𝑏 + 𝑐(𝑎 + 𝑏) + 2𝑎𝑏
= (𝑎 − 𝑏)2 + 𝑐(𝑎 + 𝑏) + 2𝑎𝑏 > 𝑐 2 + 𝑎𝑏 et cycl.
so 𝑎2 + 𝑏𝑐, 𝑏 2 + 𝑐𝑎, 𝑐 2 + 𝑎𝑏 can always form the sides of a triangle.
4(iv) Since 𝑎 + 𝑏 > 𝑎 𝑎𝑛𝑑 𝑏,
𝑓(𝑎)
>
𝑓(𝑎+𝑏)
and
𝑓(𝑏)
>
𝑓(𝑎+𝑏)
𝑎 𝑎+𝑏 𝑏 𝑎+𝑏

Since 𝑐 < 𝑎 + 𝑏, 𝑓(𝑐) < 𝑓(𝑎 + 𝑏)

𝑎𝑓(𝑎) 𝑏𝑓(𝑏)
Thus 𝑓(𝑎) + 𝑓(𝑏) > + = 𝑓(𝑎 + 𝑏) > 𝑓(𝑐) et cycl.
𝑎+𝑏 𝑎+𝑏
So 𝑓(𝑎), 𝑓(𝑏) and 𝑓(𝑐) can form the sides of a triangle.

©UCLES 2020
 5(i) 𝑥 − 𝑞(𝑥) = ∑𝑛−1 𝑟 𝑛−1 𝑛−1 𝑟
𝑟=0 𝑎𝑟 × 10 − ∑𝑟=0 𝑎𝑟 = ∑𝑟=0 𝑎𝑟 × (10 − 1)

10𝑟 ≥ 1 ∀𝑟, so 𝑥 − 𝑞(𝑥) is non-negative

9|(10𝑟 − 1) ∀𝑟
5(ii) 𝑥 − 44𝑞(𝑥) = 44(𝑥 − 𝑞(𝑥)) = 43𝑥
So it is a multiple of 9 iff 43𝑥 is.
(43,9) = 1, so 𝑥 − 44𝑞(𝑥) is a multiple of 9 iff 𝑥 is

If 𝑥 has 𝑛 digits, 𝑞(𝑥) ≤ 9𝑛

Since 𝑥 = 44𝑞(𝑥), 𝑥 ≤ 396𝑛.
Any 𝑛 digit number must be at least 10𝑛−1 .
These inequalities cannot be simultaneously true for 𝑛 ≥ 5 (396 × 5 < 104 ).
Therefore 𝑛 ≤ 4.
Since 𝑥 − 44𝑞(𝑥) = 0, which is a multiple of 9, 𝑥 is a multiple of 9.
𝑞(𝑥) is an integer and 𝑥 = 44𝑞(𝑥), so 𝑥 is a multiple of 44.
Since (9,44) = 1, 𝑥 must be a multiple of 44 × 9 = 396.
So 𝑥 = 396𝑘 and therefore (by the result above) 𝑘 ≤ 4.
Checking: Only 𝑘 = 2 works.
5(iii) 𝑥 − 107𝑞(𝑞(𝑥)) = 0 = 107(𝑥 − 𝑞(𝑥)) + 107 (𝑞(𝑥) − 𝑞(𝑞(𝑥))) − 106𝑥

(𝑥 − 𝑞(𝑥)) and (𝑞(𝑥) − 𝑞(𝑞(𝑥))) are both divisible by 9 (by part (i))
and so 𝑥 is divisible by 9
𝑥 = 107𝑞(𝑞(𝑥)) and so is divisible by 107, and so is divisible by 963.
So 𝑥 = 963𝑘 for some 𝑘.
If 𝑥 has 𝑛 digits, then 𝑞(𝑥) ≤ 9𝑛. By (i), 𝑞(𝑞(𝑥)) ≤ 𝑞(𝑥) ≤ 9𝑛.
So 𝑥 ≤ 963𝑛 and 𝑥 ≥ 10𝑛−1 which implies that 𝑛 ≤ 4 and so 𝑘 ≤ 4
Checking: Only 𝑘 = 1 works.

©UCLES 2020
 6(i)  a 2  bc b  a  d  
a b  2
Let M    ; then M   
 c  a  d  d  bc 
c d  2 

so Tr(M2) = a2 + d2 +2bc = (a + d)2 – 2(ad – bc)

6(ii) a b  2  a   b 
Let M    ; then M   c , where  = Tr(M) and  = Det(M).
 c d   d   

Thus M2 = I ⇔  = 0 and  = ∓1
or b = c = 0 and a2 = d2 = ±1
If 𝑏 = 𝑐 = 0 and 𝑎 = 𝑑 = ±1 , then M = I
If 𝑏 = 𝑐 = 0 and 𝑎 = −𝑑 = ±1, then  = 0 and  = –1
Thus M2 = +I ⇔  = 0 and  = –1.
Thus M2 = –I ⇔  = 0 and  = +1.
6(iii) Part (ii) implies Det(M2) = – 1, if M4 = I, but M2 ≠ I .
However, Det(M2) = Det(M)2 , so this is impossible.

Clearly M2 = I ⇒ M4 = I
Part (ii) implies that M4 = – I ⇔ Tr(M2) = 0 and Det(M2) = 1

so from (i) ⇔ Tr(M)2 = 2Det(M) and Det(M) = 1

so ⇔ Tr(M) =  2
and Det(M) = 1.
Any example, for instance a matrix satisfying the conditions for any of M2 = I, M2 = –I,
M4 = –I, which is not a rotation or reflection.

©UCLES 2020
 7(i) 1  ti
2
1  ti 1  ti   1
|w – 1| =2
 , which is independent of 𝑡.
1  ti 1  ti 1  ti 
Points on the line 𝑅𝑒(𝑧) = 3 have the form 𝑧 = 3 + 𝑡𝑖 and
the points satisfying |𝑤 − 1| = 1 lie on a circle with centre 1.
If z = p + ti, then
2
 2   p  2 c 
2
2 2   p  2  c  cti  c 2t 2
w c  
 p  2  ti  p  2 2  t 2
which is independent of t when  2   p  2  c   c 2  p  2 
2 2

1
which is when c  .
p2
1 1
Thus the circle has centre at and radius .
p2 p2

2 2  p  2   2ti
w  ,
 p  2   ti  p  2 2  t 2
so Im(w) > 0 when t < 0; that is, for those z on V with negative imaginary part.
7(ii) If z = t + qi then
2 2 2
2 2  cq   t  2  ci c 2  t  2    cq  2 
w  ci = 
 t  2   qi  t  2 2  q 2
2
which is independent of t when  cq  2   c 2q 2

1
which is when c  
q
1 1
so the circle has centre  i A1 and radius c2  A1.
q q

2 2  t  2   2qi
w  ,
 t  2   qi  t  2 2  q 2
so Re(w) > 0 when t > 2; that is, for those z on H with real part greater than 2.

©UCLES 2020
 8(i)

𝑎
𝑏 𝑐

Graph: Zeroes at 𝑥 = 0, 𝑐 and one other point (ℎ: label not required) in (𝑎, 𝑏).
Graph: Turning points at 𝑥 = 0, 𝑎, 𝑏, 𝑐.
Graph: Quintic shape with curve below axis in (0, ℎ) and above axis in (ℎ, 𝑐)
The area conditions give 𝐹(0) = 𝐹(𝑐) = 0.
𝐹 ′ (𝑥) = 𝑓(𝑥), so 𝐹 ′ (0) = 𝐹 ′ (𝑎) = 𝐹 ′ (𝑏) = 𝐹 ′ (𝑐) = 0
Since 𝑓 is a quartic and the coefficient of 𝑥 4 is 1,
1
𝐹 must be a quintic and the coefficient of 𝑥 5 is .
5
𝐹(0) = 𝐹 ′ (0) = 0 and 𝐹(𝑐) = 𝐹 ′ (𝑐) = 0, so 𝐹 must have double roots at 𝑥 = 0 and 𝑐.
So 𝐹(𝑥) must have the given form.
[Explanation must be clear that the double roots are deduced from the fact that
F(x) = F ′ (x) = 0 at those points.]

1
𝐹(𝑥) + 𝐹(𝑐 − 𝑥) = 𝑥 2 (𝑥 − 𝑐)2 [(𝑥 − ℎ) + (𝑐 − 𝑥 − ℎ)]
5
1
= 𝑥 2 (𝑐 − 𝑥)2 (𝑐 − 2ℎ)
5

8(ii) Let 𝐴 be the (positive) area enclosed by the curve between 0 and 𝑎.
The maximum turning point of 𝐹(𝑥) occurs at 𝑥 = 𝑏, with 𝐹(𝑏) = 𝐴.
The minimum turning point of 𝐹(𝑥) occurs at 𝑥 = 𝑎, with 𝐹(𝑎) = −𝐴.
Therefore 𝐹(𝑥) ≥ −𝐴, with equality iff 𝑥 = 𝑎.
So 𝐹(𝑏) + 𝐹(𝑥) ≥ 0, with equality iff 𝑥 = 𝑎.
𝐹(𝑎) + 𝐹(𝑥) ≤ 0, with equality iff 𝑥 = 𝑏.
1
Since 𝐹(𝑏) + 𝐹(𝑐 − 𝑏) = 𝑏 2 (𝑐 − 𝑏)2 (𝑐 − 2ℎ),
5
either 𝑐 > 2ℎ, or 𝑐 = 2ℎ and 𝑐 − 𝑏 = 𝑎.
1
Also, 𝐹(𝑎) + 𝐹(𝑐 − 𝑎) = 𝑎2 (𝑐 − 𝑎)2 (𝑐 − 2ℎ), so
5
either 𝑐 < 2ℎ, or 𝑐 = 2ℎ and 𝑐 − 𝑎 = 𝑏.
Thus 𝑐 = 𝑎 + 𝑏 and 𝑐 = 2ℎ.
8(iii) 𝐹(𝑥) =
1
𝑥 2 (𝑥 − 𝑐)2 (2𝑥 − 𝑐)
10
1
So 𝑓(𝑥) = 𝑥(𝑥 − 𝑐)(5𝑥 2 − 5𝑥𝑐 + 𝑐 2 )
5

©UCLES 2020
 The roots of the quadratic factor must be 𝑎 and 𝑏.
1
𝑓(𝑥) = (5𝑥 4 − 10𝑐𝑥 3 + 6𝑐 2 𝑥 2 − 𝑐 3 𝑥)
5
′ (𝑥) 1
𝑓 = (20𝑥 3 − 30𝑐𝑥 2 + 12𝑐 2 )
5
1 12
𝑓 ′′ (𝑥) = (60𝑥 2 − 60𝑐𝑥 + 12𝑐 2 ) = (5𝑥 2 − 5𝑐𝑥 + 𝑐 2 )
5 5

Therefore 𝑓 ′′ (𝑥) = 0 at 𝑥 = 𝑎 and 𝑥 = 𝑏 and so (𝑎, 0) and (𝑏, 0) are points of inflection.

©UCLES 2020
 9 If the particles collide at time 𝑡:
𝑉𝑡 + 𝑈𝑡 cos 𝜃 = 𝑑, and
1 1
ℎ − 𝑔𝑡 2 = 𝑈𝑡 sin 𝜃 − 𝑔𝑡 2 (or ℎ = 𝑈𝑡 sin 𝜃)
2 2

Therefore, 𝑑 sin 𝜃 − ℎ cos 𝜃 = 𝑉𝑡 sin 𝜃 + 𝑈𝑡 sin 𝜃 cos 𝜃 − 𝑈𝑡 sin 𝜃 cos 𝜃

𝑉ℎ
=
𝑈

9(i) Dividing the previous result by 𝑑 cos 𝜃 gives:

ℎ 𝑉ℎ
tan 𝜃 − = >0
𝑑 𝑈𝑑 cos 𝜃
ℎ
Since tan 𝛽 = , tan 𝜃 > tan 𝛽 and so 𝜃 > 𝛽
𝑑

9(ii) The height of collision must be non-negative, so 𝑈𝑡 sin 𝜃 − 𝑔𝑡 2 ≥ 0.

1
2
1 1 ℎ 𝑔ℎ
So 𝑈 sin 𝜃 ≥ 𝑔𝑡 = 𝑔 or (𝑈 sin 𝜃)2 ≥
2 2 𝑈 sin 𝜃 2
𝑔ℎ
Therefore 𝑈 sin 𝜃 ≥ √ .
2

9(iii) 𝑑 sin 𝜃 − ℎ cos 𝜃 can be written as √𝑑 2 + ℎ2 sin(𝜃 − 𝛽)

So 𝑑 sin 𝜃 − ℎ cos 𝜃 < √𝑑 2 + ℎ2 (since 𝜃 > 𝛽)
𝑉ℎ
Therefore, < √𝑑 2 + ℎ2
𝑈
ℎ 𝑈
or sin 𝛽 = <
√𝑑 2 +ℎ2 𝑉

The height at which the particles collide is:

1 𝑔ℎ2
ℎ − 𝑔𝑡 2 = ℎ −
2 2𝑈 2 sin2 𝜃
𝑔ℎ2 1
ℎ− > ℎ iff 𝑈 2 sin2 𝜃 > 𝑔ℎ
2𝑈 2 sin2 𝜃 2

The vertical velocity of the particle fired from 𝐵 at the point of collision is:
𝑔ℎ
𝑈 sin 𝜃 − 𝑔𝑡 = 𝑈 sin 𝜃 −
𝑈 sin 𝜃
𝑔ℎ
𝑈 sin 𝜃 − > 0 iff 𝑈 2 sin2 𝜃 > 𝑔ℎ
𝑈 sin 𝜃

Since both cases have the same condition:

1
The particles collide at a height greater than ℎ if and only if the particle projected from 𝐵
2
is moving upwards at the time of collision.

©UCLES 2020
 10(i)
H

R 𝑇

L P

2𝜃
𝑚𝑔

Diagram showing necessary forces and angles

𝜆(2𝑎 cos 𝛼−𝑙)
𝑇=
𝑙

Resolving tangentially:
𝑇 sin 𝛼 − 𝑚𝑔 sin 2𝛼 = 0
𝜆
Therefore sin 𝛼 ( (2𝑎 cos 𝛼 − 𝑙) − 2𝑚𝑔 cos 𝛼) = 0
𝑙

Since sin 𝛼 > 0, 2𝑎𝜆 cos 𝛼 − 𝜆𝑙 − 2𝑚𝑔𝑙 cos 𝛼 = 0

𝜆𝑙
cos 𝛼 =
2(𝑎𝜆−𝑚𝑔𝑙)

cos 𝛼 < 1, so 𝜆𝑙 < 2(𝑎𝜆 − 𝑚𝑔𝑙)

Therefore 𝜆(2𝑎 − 𝑙) > 2𝑚𝑔𝑙
2𝑚𝑔𝑙
Since 2𝑎 − 𝑙 > 0, 𝜆 >
2𝑎−𝑙

10(ii) Energy:
1 𝜆 1 𝜆
𝑚𝑣 2 − 𝑚𝑔𝑎 cos 2𝜃 + (2𝑎 cos 𝜃 − 𝑙)2 = 𝑚𝑢2 − 𝑚𝑔𝑎 + (2𝑎 − 𝑙)2
2 2𝑙 2 2𝑙

If the particle comes to rest when 𝜃 = 𝛽:

𝜆 1 𝜆
−𝑚𝑔𝑎 (2cos 2 𝛽 − 1) + (2𝑎 cos 𝛽 − 𝑙)2 = 𝑚𝑢2 − 𝑚𝑔𝑎 + (2𝑎 − 𝑙)2
2𝑙 2 2𝑙
2(𝑎𝜆−𝑚𝑔𝑙) 1 2𝜆𝑎2
aλcos 2 𝛽 ( ) − 2𝑎𝜆 cos 𝛽 = 𝑚𝑢2 − 2𝑚𝑔𝑎 + − 2𝑎𝜆
𝜆𝑙 2 𝑙
𝑚𝑢2
Therefore, cos 2 𝛽 − 2 cos 𝛼 cos 𝛽 = cos 𝛼 + 1 − 2 cos 𝛼
2𝑎𝜆

Adding cos 2 𝛼 to both sides:

𝑚𝑢2
(cos 𝛼 − cos 𝛽)2 = (1 − cos 𝛼)2 + cos 𝛼
2𝑎𝜆

For this to occur, cos 𝛽 > 0:

𝑚𝑢2
cos 2 𝛼 > (1 − cos 𝛼)2 + cos 𝛼
2𝑎𝜆
2𝑎𝜆
And so, 𝑢2 < (2 − sec 𝛼)
𝑚

©UCLES 2020
 11(i) If the game has not ended after 2𝑛 turns, then the sequence has either been 𝑛 repetitions
of 𝐻𝑇 or 𝑛 repetitions of 𝑇𝐻.
So 𝑃(𝐺𝑎𝑚𝑒 ℎ𝑎𝑠 𝑛𝑜𝑡 𝑓𝑖𝑛𝑖𝑠ℎ𝑒𝑑 𝑎𝑓𝑡𝑒𝑟 2𝑛 𝑡𝑢𝑟𝑛𝑠) = 2(𝑝𝑞)𝑛 .
So the probability that the game never ends is lim𝑛→∞ 2(𝑝𝑞)𝑛 = 0.
Sequence that follows the first 𝐻 will be 𝑘 repetitions of 𝑇𝐻, followed by 𝐻, where 𝑘 ≥ 0.
𝑝
So 𝑃(𝐴 𝑤𝑖𝑛𝑠|𝑓𝑖𝑟𝑠𝑡 𝑡𝑜𝑠𝑠 𝑖𝑠 𝐻) = ∑∞ 𝑘
𝑘=0(𝑝𝑞) 𝑝 = 1−𝑝𝑞
𝑝
𝑃(𝐴 𝑤𝑖𝑛𝑠 ∩ 𝑓𝑖𝑟𝑠𝑡 𝑡𝑜𝑠𝑠 𝑖𝑠 𝐻) = 𝑝 ×
1−𝑝𝑞

If first toss is a tail then the sequence that follows would be

𝑘 repetitions of 𝐻𝑇 followed by 𝐻𝐻.
𝑝2
So 𝑃(𝐴 𝑤𝑖𝑛𝑠|𝑓𝑖𝑟𝑠𝑡 𝑡𝑜𝑠𝑠 𝑖𝑠 𝑇) =
1−𝑝𝑞

𝑝2 𝑞
𝑃(𝐴 𝑤𝑖𝑛𝑠 ∩ 𝑓𝑖𝑟𝑠𝑡 𝑡𝑜𝑠𝑠 𝑖𝑠 𝑇) =
1−𝑝𝑞

𝑝2 (1+𝑞)
Therefore 𝑃(𝐴 𝑤𝑖𝑛𝑠) =
1−𝑝𝑞

11(ii) Following a first toss of 𝐻:

𝐴 wins with 𝐻𝐻
or (𝐻𝑇 followed by any sequence where A wins after first toss was 𝑇)
or (𝑇 followed by any sequence where A wins after first toss was 𝑇)
The probabilities of these cases are:
𝑝2
𝑝𝑞 𝑃(𝐴 wins |the first toss is a tail)
𝑞 𝑃(𝐴 wins |the first toss is a tail)
Therefore:
𝑃(A wins | the first toss is a head) = 𝑝2 + (𝑞 + 𝑝𝑞)𝑃(A wins | the first toss is a tail)

©UCLES 2020
 Similarly, following first toss of 𝑇:
𝐴 wins with (𝐻 followed by any sequence where A wins after first toss was H)
or (𝑇𝐻 followed by any sequence where A wins after first toss was 𝐻)
Therefore:
𝑃(A wins | the first toss is a tail) = (𝑝 + 𝑝𝑞)𝑃(A wins | the first toss is a head)
So
𝑃(𝐴 | 𝐻 𝑓𝑖𝑟𝑠𝑡) = 𝑝2 + (𝑞 + 𝑝𝑞)(𝑝 + 𝑝𝑞)𝑃(𝐴 | 𝐻 𝑓𝑖𝑟𝑠𝑡)
𝑝2
𝑃(𝐴 | 𝐻 𝑓𝑖𝑟𝑠𝑡) =
1−(𝑝+𝑝𝑞)(𝑞+𝑝𝑞)

And
𝑃(𝐴 | 𝑇 𝑓𝑖𝑟𝑠𝑡) = (𝑝 + 𝑝𝑞)(𝑝2 + (𝑞 + 𝑝𝑞)𝑃(𝐴 | 𝑇 𝑓𝑖𝑟𝑠𝑡))
𝑝2 (𝑝+𝑝𝑞)
𝑃(𝐴 | 𝑇 𝑓𝑖𝑟𝑠𝑡) =
1−(𝑝+𝑝𝑞)(𝑞+𝑝𝑞)

So
𝑝2 𝑝2 (𝑝+𝑝𝑞) 𝑝2 (1−𝑞 3 )
𝑃(𝐴 𝑤𝑖𝑛𝑠) = 𝑝 × +𝑞× =
1−(𝑝+𝑝𝑞)(𝑞+𝑝𝑞) 1−(𝑝+𝑝𝑞)(𝑞+𝑝𝑞) 1−(1−𝑝2 )(1−𝑞 2 )

11(iii) Let 𝑊 be the event that 𝐴 wins the game.

𝑃(𝑊 | 𝐻 𝑓𝑖𝑟𝑠𝑡) = 𝑝𝑎−1 + (1 + 𝑝 + 𝑝2 + ⋯ + 𝑝𝑎−2 )𝑞𝑃(𝑊 | 𝑇 𝑓𝑖𝑟𝑠𝑡)
𝑃(𝑊 | 𝑇 𝑓𝑖𝑟𝑠𝑡) = (1 + 𝑞 + 𝑞 2 + ⋯ + 𝑞 𝑏−2 )𝑝𝑃(𝑊 | 𝐻 𝑓𝑖𝑟𝑠𝑡)
𝑝𝑎−1
𝑃(𝑊 | 𝐻 𝑓𝑖𝑟𝑠𝑡) =
1−(1−𝑝𝑎−1 )(1−𝑞 𝑏−1 )

𝑝𝑎−1 (1−𝑞 𝑏−1 )

𝑃(𝑊 | 𝑇 𝑓𝑖𝑟𝑠𝑡) =
1−(1−𝑝𝑎−1 )(1−𝑞 𝑏−1 )

Therefore:
𝑝𝑎−1 (1−𝑞 𝑏 )
𝑃(𝑊) =
1−(1−𝑝𝑎−1 )(1−𝑞 𝑏−1 )

If 𝑎 = 𝑏 = 2,
𝑝(1−𝑞 2 ) 𝑝2 (1+𝑞)
𝑃(𝑊) = = as expected.
1−(1−𝑝)(1−𝑞) 1−𝑝𝑞

©UCLES 2020
 12(i) For the biased die:
1 2
𝑃(𝑅1 = 𝑅2 ) = ∑𝑛𝑖=1 ( + 𝜀𝑖 )
𝑛
1 2
𝑃(𝑅1 = 𝑅2 ) = ∑𝑛𝑖=1 1 + ∑𝑛𝑖=1 𝜀𝑖 + ∑𝑛𝑖=1 𝜀𝑖 2
𝑛2 𝑛

∑𝑛𝑖=1 𝜀𝑖 = 0 , so
1
𝑃(𝑅1 = 𝑅2 ) = + ∑𝑛𝑖=1 𝜀𝑖 2
𝑛
1
For a fair die, 𝑃(𝑅1 = 𝑅2 ) = and ∑𝑛𝑖=1 𝜀𝑖 2 > 0, so it is more likely with the biased die.
𝑛

(ii) 1
𝑃(𝑅1 > 𝑅2 ) = (1 − 𝑃(𝑅1 = 𝑅2 ))
2

Therefore, the value of 𝑃(𝑅1 > 𝑅2 ) if the die is possibly biased is ≤ 𝑃(𝑅1 > 𝑅2 ) if the die
is fair.
𝑥
Let T= ∑𝑛𝑟=1 𝑥𝑟 and, for each 𝑖, let 𝑝𝑖 = 𝑖
𝑇
Then ∑𝑛𝑖=1 𝑝𝑖 = 1, so we can construct a biased 𝑛-sided die with 𝑃(𝑋 = 𝑖) = 𝑝𝑖
𝑃(𝑅1 > 𝑅2 ) = ∑𝑛𝑖=2 ∑𝑖−1
𝑗=1 𝑝𝑖 𝑝𝑗

For a fair die:

𝑛−1
𝑃(𝑅1 > 𝑅2 ) =
2𝑛

Therefore
𝑥𝑖 𝑥𝑗 𝑛−1
∑𝑛𝑖=2 ∑𝑖−1
𝑗=1 2 ≤ and so
𝑇 2𝑛
𝑛 𝑖−1 𝑛 2
𝑛−1
∑ ∑ 𝑥𝑖 𝑥𝑗 ≤ (∑ 𝑥𝑖 )
2𝑛
𝑖=2 𝑗=1 𝑖=1

(iii) For the biased die:

1 3
𝑃(𝑅1 = 𝑅2 = 𝑅3 ) = ∑𝑛𝑖=1 ( + 𝜀𝑖 )
𝑛
1 3𝜀𝑖 3𝜀𝑖2
= ∑𝑛𝑖=1 + ∑𝑛𝑖=1 + ∑𝑛𝑖=1 + ∑𝑛𝑖=1 𝜀𝑖 3
𝑛3 𝑛2 𝑛

Therefore
𝑛 𝑛 𝑛
3𝜀𝑖 3𝜀𝑖2
𝑃(𝑅1 = 𝑅2 = 𝑅3 𝑏𝑖𝑎𝑠𝑒𝑑) − 𝑃(𝑅1 = 𝑅2 = 𝑅3 𝑓𝑎𝑖𝑟) = ∑ 2 + ∑ + ∑ 𝜀𝑖 3
𝑛 𝑛
𝑖=1 𝑖=1 𝑖=1
2
3𝜀
= ∑𝑛𝑖=1 𝑖 + ∑𝑛𝑖=1 𝜀𝑖 3 (since ∑𝑛𝑖=1 𝜀𝑖 = 0)
𝑛
3𝜀𝑖2 3
= ∑𝑛𝑖=1 + 𝜀𝑖3 = ∑𝑛𝑖=1 𝜀𝑖2 ( + 𝜀𝑖 )
𝑛 𝑛
1
But 𝜀𝑖 ≥ − (since 𝑝𝑖 ≥ 0), so this sum must be positive.
𝑛

Therefore, 𝑃(𝑅1 = 𝑅2 = 𝑅3 ) must be greater for the biased die.

©UCLES 2020
Cambridge Assessment Admissions Testing offers a range of tests to support selection and
recruitment for higher education, professional organisations and governments around the world.
Underpinned by robust and rigorous research, our assessments include:

 assessments in thinking skills

 admissions tests for medicine and healthcare
 behavioural styles assessment
 subject-specific admissions tests.

We are part of a not-for-profit department of the University of Cambridge.

Cambridge Assessment
Admissions Testing
The Triangle Building
Shaftesbury Road
Cambridge
CB2 8EA
United Kingdom

Admissions tests support:

www.admissionstestingservice.org/help

©UCLES 2020