FOLIA 345

Annales Universitatis Paedagogicae Cracoviensis

Studia ad Didacticam Mathematicae Pertinentia 13(2021)
ISSN 2080-9751
DOI 10.24917/20809751.13.6

Barbara Barańska, Małgorzata Zambrowska

What do we mean by mean?∗

Abstract. In order to be successful with a certain concept at school, it is

often sufficient to master some types of tasks, without having to be aware
of the different perspectives that a concept can be viewed from. It is, how-
ever, definitely not enough when the goal is to understand mathematics and
become flexible and creative in solving mathematical tasks. Not only school
students, but also prospective teachers should be encouraged to reflect on
what particular concepts actually mean and how they can be thought of.
In this article we focus our attention on the concept of mean. In the
introduction we outline the areas where our contribution adds to the results
of the research carried out so far. Then we provide a brief review of how the
Polish core curriculum and two very popular mathematics textbook series
(one each for primary and secondary school) treat the topic of means. Next,
we present results from a study conducted on a group of first-year students of
mathematics and a group of teachers undertaking postgraduate studies qual-
ifying for teaching mathematics as a second subject. The study was aimed
at diagnosing pre-service teachers’ substantive competences in mathematics.
In this paper we show only the results obtained in two tasks concerning the
arithmetic and geometric means. The paper ends with several suggestions
and recommendations which can be useful for and applied by teachers of
mathematics as well as mathematics teachers’ educators.

1. Introduction
According to Bakker and Gravemeijer (2006)

the term ”average” itself stems from maritime law, in connection with
insurance and the fair share of profit and loss. The Oxford English Dic-
tionary (Simpson and Weiner, 1989), under the heading of ”average,”
states that one of the meanings of average in maritime law is ”the equi-
table distribution of expense or loss, when of general incidence, among
∗ 2020
Mathematics Subject Classification:97B50
Keywords and phrases: mathematics teacher education; means
[6] Barbara Barańska, Małgorzata Zambrowska

all the parties interested, in proportion to their several interests.” More

generally, the average came to mean the distribution of the aggregate
inequalities (in quantity, quality, intensity, etc.) of a series of things
among all the members of the series, so as to equalize them. In its
transferred use the term average thus came to signify the arithmetic
mean (Bakker and Gravemeijer, 2006, p. 153).

Already centuries ago, intuitions related to the concept of the arithmetic mean
were used to make estimations and solve practical problems (Bakker, 2003). But
in the times of Pythagoras there were also other two known mean values: the
harmonic and geometric mean (Heath, 1981). However, as noted by Landtblom
(2018), in the traditional teaching of averages, the notion of arithmetic mean has
received more attention than other statistical concepts. It seems important for
researchers to pay more attention also to other averages since they are taught at
school.
Several authors have made a great effort to characterize in details some of the
means for educational purposes. For instance, Strauss and Bichler (1988) found
and highlighted seven important features of the arithmetic mean (the average in
the passage below):

A. The average is located between the extreme values.

B. The sum of the deviations from the average is zero.
C. The average is influenced by values other than the average.
D. The average does not necessarily equal one of the values that was
summed.
E. The average can be a fraction that has no counterpart in physical
reality.
F. When one calculates the average, a value of zero, if it appears, must
be taken into account.
G. The average value is representative of the values that were averaged
(Strauss and Bichler, 1988, p. 66; see also: Leon and Zawojewski, 1990).

Bakker and Gravemeijer (2006) point out to some other aspects of mean that
they revealed when looking at this concept through the phenomenological lenses.
They consider the following elements important:

the implicit use of the mean to estimate large numbers, the strategy
of calculating the mean as a means to eliminate measurement errors,
the change of perspective by which the mean becomes an entity in and
of itself, the role of mean-as-a-measure, and the difference between the
natural sciences (where there seems to be an implicit expectation of
symmetrical distributions) and the social sciences (where the appar-
ent skewed distributions create the need to describe those distributions,
which creates a role for the median). (Bakker and Gravemeijer, 2006,
p. 161).
What do we mean by mean? [7]

School students associate the concept of average with various notions e.g., the
sum of all numbers considered, a half, the average of the smallest and largest
number, the sum of numbers divided by the number of these numbers (Bakker,
2003). Mokros and Russell (1995) found five predominant approaches to solving
problems involving averages. In the group they studied, the students thought of
an average as mode, an algorithmic procedure, a tool for making sense of the data
set, midpoint or a mathematical point of balance. As research shows students can
be able to recognize the algorithm for a particular mean, calculate the mean with
the use of appropriate algorithm, or even find a new mean when some number is
added to the examined set of numbers, but it may be more difficult to them to
find an unknown value of a number when given the other numbers and the mean
of the whole set (Cai, 1995; Guimarães et al., 2010). Also the students often lack
understanding of the arithmetic mean as a representative value of a set of numbers,
which has been found to be the most difficult property of this mean (Strauss and
Bichler, 1988).
A significant problem is that students may be able to calculate the average of
a given set of numbers, and yet remain unaware of its certain qualitative features.
Leavy and O’Loughlin (2006), for instance, distinguish between computational and
conceptual interpretation of the arithmetic mean:

Computationally, the arithmetic average is the score around which de-

viations in one direction exactly equal deviations in another direction.
Conceptually, the mean acts as a mathematical point of balance in the
distribution similar to a center of gravity, and is considered represen-
tative of a data set. (Leavy and O’Loughlin, 2006, p.55)

Knowing how to determine a given mean is not an indication of understanding

of the concept (Batanero et al., 1994; Watson and Moritz, 2000). Students may
experience difficulties with means at both the procedural level and the level of
conceptual understanding. Teaching which is dominated by the use of algorithms
(e.g., Gfeller, Niess and Lederman, 1999), can lead students to become helpless in
the face of problems requiring the use of conceptual understanding of means.
Surprisingly, or perhaps not, prospective teachers happen to experience very
similar difficulties in understanding mathematical concepts as the students (Ja-
cobbe and Carvalho, 2011). Even recently (Landtblom, 2018), it has been con-
cluded that there is still too little research available that advances our knowledge
of pre- and in-service teachers’ understanding of means.
There are four gaps which this paper addresses to some extent. First, we are
not aware of any recent Polish educational study on mathematics, that has looked
at issues related to the concept of means. Secondly, in the piece of research that
we report, we show how some teachers taking postgraduate studies qualifying for
teaching mathematics deal with arithmetic and geometric mean tasks, and this
group of prospective teachers has been underreported in research to date. Thirdly,
while among the different means, the arithmetic mean seems to attract the most
attention from researchers, we also address the problem of the geometric mean
and mention the harmonic mean. Last, but not least, at the end of this paper,
we propose a certain universal – yet not found in contemporary Polish school
[8] Barbara Barańska, Małgorzata Zambrowska

textbooks – perspective on the three means, which helps to unify the way we talk
and think of them.

2. The treatment of means in the Polish core curriculum and

mathematics school textbooks
2.1. Means in the core curriculum
In the core curriculum for primary school (grades 4-8), in the section entitled
”Reading data and elements of descriptive statistics”, it is stated that the student
is to calculate the arithmetic mean of several numbers. In the section ”Conditions
and method of implementation”, which contains extended comments of the authors
of the document, there is an example of a task that a primary school student should
be able to solve. It reads as follows:
Maciek got 10 marks in mathematics. Here are 9 of them: 2, 2, 2, 3,
3, 4, 5, 5, 6. The arithmetic mean of all his 10 marks is equal to 3.6 .
Determine the missing grade.
In the core curriculum for secondary schools (4-year secondary school or 5-
year technical school), the topic of averages appears in the section ”Probability
and statistics”. It says that a student who pursues mathematics education at the
basic level ”calculates the arithmetic mean and the weighted mean”.
It is important to notice, however, that these averages (and others not specified
above) are also used by high school students many times on the occasion of other
topics, not explicitly related to the average, and this is not emphasized in the
curriculum. The arithmetic mean, for instance, is used to determine the coordinates
of the vertex of a parabola (being a graph of a quadratic function), when the
coordinates of two points of the parabola, symmetrically located with respect to
its main axis, are known. Also, in an arithmetic sequence, of every three consecutive
terms of the sequence, the middle term is the arithmetic mean of the neighboring
terms. The weighted average is used by the students when they calculate the
average of their grades, e.g. from tests, which are assigned different weights. In
a geometric sequence with positive terms, out of every three consecutive terms of
the sequence, the middle term is the geometric mean of the neighboring terms.
Moreover, the length of the height subtended from the vertex of a right angle in
a triangle is the geometric mean of the lengths of the segments into which it divides
the hypotenuse.
A trapezoid is an example of a geometrical figure such that one can see four
different means in it1 :
• the length of the segment connecting the centers of the arms of a trapezoid
is the arithmetic mean of the lengths of its bases,
• the length of the segment parallel to the trapezoid bases and dividing this
trapezoid into two similar trapezoids is the geometric mean of the lengths of
the trapezoid bases,
1 Another example of such a figure could be the semi-circle (Hess, 1961; see also:
http://www.eudoxos.pl/srednie/).
What do we mean by mean? [9]

• the length of the segment parallel to the trapezoid bases and passing through
the point of intersection of its diagonals is the harmonic mean of the lengths
of the trapezoid bases,
• the length of the segment parallel to the bases of the trapezoid and dividing
it into two trapezoids of equal areas is the quadratic mean of the lengths of
the trapezoid bases.

2.2. Arithmetic mean in a popular series of Polish primary school mathematics

textbooks
In the series of textbooks for primary school under review2 , the concept of
arithmetic mean appears for the first time in grade 5. The students first calcu-
late the arithmetic mean of several numbers while discussing decimal fractions,
and then return to this concept and practice the skills they have acquired when
discussing whole numbers. The concept is introduced as follows:
The arithmetic mean of two numbers is the quotient of the sum of those
numbers by 2.
The arithmetic mean of the numbers 3 and 5.5:
3 + 5, 5 8, 5
= = 4, 25
2 2
The fractional dash replaces the division sign.
The arithmetic mean of three numbers is the quotient of the sum of
those numbers by 3.
The arithmetic mean of the numbers 6, 7.3 and 8:
6 + 7, 3 + 8 21, 3
= = 7, 1
3 3
The arithmetic mean of four numbers is the quotient of the sum of
those numbers by 4.
And so on (5th grade textbook, p. 166).
The task which precedes this introduction of the arithmetic mean concerns a ream
of paper (500 sheets), whose thickness and weight are given. The students’ task
is to calculate the thickness and weight of one sheet of paper. This task requires
students to consider a fair share that would take place if the thickness and weight
of all the sheets were distributed equally. Immediately after the introduction of
the arithmetic mean, the students calculate arithmetic means of two, three and
four decimal numbers given and in a next task they calculate the average of some
girl’s grades. The topic of means returns when the whole numbers are discussed.
Among other things, students calculate the average air temperature on the basis
of temperature measurements taken on several consecutive days.
2 We chose the textbooks published by Gdańskie Wydawnictwo Oświatowe, as they are very

(if not the most) popular among mathematics teachers and widely used at primary schools in
Poland.
[10] Barbara Barańska, Małgorzata Zambrowska

In the 6th grade textbook, there are tasks requiring the determination of the
arithmetic mean, but the topic itself is not explicitly addressed.
In grade 7, the students revisit the concept of the arithmetic mean, they know
how to calculate the arithmetic mean and also how to solve a text problem related
to the mean. This concept comes at the end of the school year, in the ”Statistics”
section. Only in this textbook in the analysed series, a separate chapter is devoted
to the concept of arithmetic mean. It begins with two tables showing the final
grades in mathematics in two 7th grade classes, with 20 and 25 students respec-
tively. Then the mean of the grades in each class is calculated separately and also
the mean of the grades in both classes together. The authors of the textbook draw
students’ attention to the fact that the mean of the two initially calculated means
is not the mean sought after in the final task.
In this chapter, the authors of the textbook offer a number of tasks. Some of
them require only the calculation of the arithmetic mean of a set of numbers placed
in a realistic context. But there are also tasks which require more than just the
application of the learned method of calculating the average. For instance: Jurek
has four grades in mathematics, we know the average of his grades, we ask by how
much this average will increase if the boy gets a six (Polish counterpart for A+);
or: Zuzanna has 12 grades on her certificate, they are only grades 4 and 5 (B’s
and A’s respectively), we know the average of her grades, we ask how many 5’s she
has. However, in our opinion, the following three tasks deserve special attention:
Task 4, page 300
Ala and her 21 classmates were writing an English test. A maximum
of 44 points could be scored. The average score was 28. No one wrote
below the average. How many points did Ala get?
Task 7, page 300
Anne keeps all her books on six shelves. There are 20 books on one shelf
on average. Determine how the average number of books on a shelf will
change when Anne:
(a) moves 10 books from the highest shelf to the lowest
b) adds 1 more book to each shelf,
c) buys 12 more books and puts them all on one of the shelves,
1
d) sells 4 of all her books.
Task 13, page 301
Each of ten glasses contains an average of s milliliters of syrup. Write
an algebraic expression that represents how much syrup on average will
be in each glass when:
a) we add 40ml of syrup to each glass,
b) we pour out half of what is in each glass,
c) we pour 100ml of syrup into one glass,
d) we pour another 100ml of syrup into each of three glasses,
What do we mean by mean? [11]

e) to the ten glasses we add a glass containing 150ml of syrup,

f) we add an empty glass to the ten glasses.

The first of these tasks, contains a statement that may initially cause consternation
in students: ”no one wrote below the mean”. This information forces students to
think. It is not enough to know the formula for the arithmetic mean of several
numbers, the students need to be able to interpret the information about the
mean well. Here, the quoted passage obviously indicates that, in that case, no
one wrote above the average either, and therefore everyone, including Ala, got the
same result.
The next two tasks can greatly contribute to the development of conceptual
understanding of the arithmetic mean in students. They allow to observe and
analyse how the arithmetic mean is affected by a change made to a set of numbers
or its parts. The students can discover and convince themselves that:

• the average of the numbers will not change if we decrease one number and
increase another number by the same amount,

• the average will increase (decrease) by k if we increase (decrease) each num-

ber by k,
k
• the average of n numbers will increase (decrease) by n if we increase (de-
crease) any of the numbers by k,

• if to the set of n numbers whose sum is S we add an n + 1th number with

value k, then the average changes from Sn to S+k S
n+1 (if k = n the change
equals 0, since the arithmetic mean remains the same, which is yet another,
interesting discovery the students can make), and so forth.

In all three examples it is not the procedure for determining the average that
is important, but the conceptual understanding of it.

2.3. Four means in a popular series of Polish secondary school mathematics

textbooks
Textbook definitions
In a 1th grade of secondary school mathematics textbook that we chose to
examine3 , a separate subsection in the chapter on algebraic expressions is devoted
to the means. The authors inform the students that a mean of a group of numbers
allows to characterize this group in a simple way and that it is contained between
the smallest and the largest number from the set. The authors introduce four types
of means:
3 We chose the textbooks published by Oficyna Edukacyjna Pazdro, as they are very (if not

the most) popular among mathematics teachers and widely used at high schools in Poland. The
topic of means is discussed in the first grade. Due to the education reform and the replacement
of former textbooks scheduled for several years, the textbooks for the fourth grade are not yet
available. This is why we do not know whether the topic will return there when statistics is
discussed and whether other means, such as quadratic, will be mentioned.
[12] Barbara Barańska, Małgorzata Zambrowska

1
1) Arithmetic mean of numbers a1 , a2 , . . . , an equals Sa = n (a1 + a2 +
· · · + an ).
2) Geometric mean of positive numbers a1 , a2 , . . . , an equals
√
Sg = a1 · a2 · · · · · an .
The following relationship holds between the geometric mean and the
arithmetic mean: if a1 , a2 , . . . , an are positive numbers then Sg ≤ Sa ,
with equality occurring only if a1 = a2 = · · · = an .
3) The weighted mean of the numbers a1 , a2 , . . . , an with positive weights
w1 , w2 , . . . , wn is equal to
a1 w1 + a2 w2 + · · · + an wn
Sw =
w1 + w2 + w3

4) The harmonic mean of positive numbers a1 , a2 , . . . , an is equal to

n
Sh = 1 1 1
a1 + a2 + an

Between the geometric mean and the harmonic mean of the n posi-
tive numbers a1 , a2 , . . . , an , the following relation holds: Sg ≥ Sh , with
equality occurring only when a1 = a2 = · · · = an (OE Pazdro, p. 132).

It is worth noting that:

• In the case of the arithmetic mean, the authors do not give any conditions
regarding the numbers for which the mean is determined. The learner may
easily guess that any real values of numbers are allowed.

• Similarly, in case of the weighted mean no conditions are given concerning

the numbers themselves, but it is emphasized that the weights take positive
values. Here it is quite easy to conclude that negative weights would not
make much sense. But what if we allow 0 as a weight? What would happen
if a number received a weight of 0? Such number would not be taken into
account (would not matter), so allowing for a zero weight would also make
no sense4 .

• Geometric and harmonic means are calculated only for positive numbers.
In the case of the harmonic mean, it is obvious that none of the numbers
can be zero, for when calculating this mean we use inverses of the numbers.
However, why is zero not allowed in the case of the geometric mean? The
root of zero is well defined after all. The authors do not comment on this in
any way. Moreover, the literature on this matter is inconsistent, since other
authors allow the numbers whose geometric mean is calculated to have zero
values (e.g., Cewe, Nahorska and Pancer, 2003).
4 However, a clear distinction must be made between a situation when the weight would be 0

and a situation when one of the numbers in the set is 0. In the latter, of course, the number 0
will be counted in the calculation of the average.
What do we mean by mean? [13]

• The exclusion of negative numbers in the case of both last mentioned means
is also left without a comment. Here, in turn, while it is easy to see that not
in every case it would be possible to determine the geometric mean in the set
of real numbers, one does not immediately see the reason why it would not
be possible to calculate the harmonic mean. If we allowed negative numbers,
for example, for the set of numbers -3,-2, 2, and 3 we would get zero in
the denominator of the fractional expression for the harmonic mean. Is this
the reason why negative numbers were eliminated? The authors say nothing
about this.
• Nowhere does the textbook also say that if we have one number, it is its own
arithmetic, geometric and harmonic mean at the same time. This trivial case,
perhaps in the authors’ opinion not even worth considering, allows
√ for, among
other things, a discussion of the degree of the arithmetic root n a. Usually,
textbook authors take the view that n is a natural number at least equal 2.
Meanwhile, Chronowski (1999) also allows for a root of degree one stating
that it is simply the same number (the root ”does nothing”). Chronowski’s
approach makes the notion of a geometric mean also meaningfully applicable
to a single number.

It should also be added that the quadratic mean, which students are going to
use (perhaps unknowingly) in the future to determine the standard deviation, is
not considered in the 1st grade textbook. For the sake of completeness, let us add
that the quadratic mean is defined5 as follows (e.g., Cewe, Nahorska and Pancer,
2003):

Quadratic For a,q

b>0 For a1q
, . . . , an > 0
mean 2
Sq = a +b
2
Sq =
a21 +a22 +···+a2n
2 n
(the root of the arithmetic mean of the
squares of numbers a1 , . . . , an )

Textbook examples
In the first example provided by the authors, the results of temperature mea-
surements from several consecutive days are given and it is shown how the average
temperature for the analyzed period was determined.
The second example states that in a certain workplace, which employed 19
workers, the average salary was 2650 PLN. After paying a new employee, the
average salary increased by 2%. It is shown how to calculate the salary of a new
employee. Both of these examples deal with the arithmetic mean.
In the third example, the following task is given:
In a clothing store, the price of a certain blouse increased first by 50%,
then by 20%. At the end of the season the price of the blouse has been
reduced by 30%. What was the average percentage change in the price
of the blouse? (OE Pazdro, p. 133)
5 Again a question arises, why do we exclude 0 from the set of the numbers whose quadratic
mean is to be found?
[14] Barbara Barańska, Małgorzata Zambrowska

In the solution given by the authors, the initial price of the blouse is denoted as
a. The final price was obviously 0, 7 · 1, 2 · 1, 5a. The authors state:
Let p denote the average factor of change in the price of the blouse.
Since there were three changes in the price of the blouse, we get the
equation
p3 a = 0, 7 · 1, 2 · 1, 5a (. . . )
p = 1, 080082 . . .
We can say that at each price change, the blouse became more expensive
by about 8% on average.
Example four shows how to calculate the weighted average of two students’ grades.
The example illustrates that although two students may have the same grades,
the weighted averages of their grades may differ if they earned different grades for
activities to which the teacher assigned different weights.
The last example contrasts two situations that are often confused by the stu-
dents. The task shown in this example is the following:
A car drove a distance from city A to city B. At what average speed
did the car travel if:
(a) it traveled half of the time with speed v1 and the other half of the
time with speed v2 ,
b) it travelled half of the distance with speed v1 and the other half of
the distance with speed v2 .
In the first case, the average speed of the car over the entire distance is equal to
the arithmetic mean of speeds v1 and v2 . The main role of this example, however,
is to show the students that when an object travels distance s at speed v1 and
then the same distance at speed v2 , its average speed over the entire route is equal
to the harmonic mean of speeds v1 and v2 , and not the arithmetic mean as one
may mistakenly guess.
Students are often familiar with the formulas for different means. That is
just part of superficial procedural knowledge (Star, 2005) on means. A deeper
understanding of the means allows, for example, not to calculate them from the
formulas when they can be determined otherwise. Unfortunately in the examined
high school textbook there are no tasks analogous to the examples we highlighted
in the primary school textbook (How would the mean change if we...) that would
stimulate out-of-the-box thinking.
Of the means students learn in high school, the geometric and harmonic means
seem to be the least intuitive. While each can be calculated for a given set of pos-
itive numbers, there is no clear message from the authors conveying the meaning
of the values obtained as means. In a task where we are not told which mean we
should use (like in the third and the last textbook examples), those who know the
meaning of each mean, should resist being deceived. The geometric mean seems to
be particularly interesting in this respect. Even just formulating the task clearly
can be a challenge in the case of this mean. On the other hand, in typical, unam-
biguously formulated tasks where the students are expected to apply the formula
What do we mean by mean? [15]

for the harmonic mean, it is usually possible to solve the task without using for-
mulas, and even without being aware that one is calculating one of the means.

3. Investigating the substantive competences of pre-service teach-

ers of mathematics - tasks on arithmetic and geometric mean
3.1. The study and its objectives
This study has been conducted as part of a larger project supported by the
National Science Center in Poland. The project’s main objective was to diagnose
the substantive and didactical competencies of pre-service teachers of mathematics,
mainly those participating in postgraduate qualifying courses. The pilot study
however has been conducted on a number of undergraduate students.
With respect to what we present in this paper, we were first interested in
whether the study participants could correctly solve the tasks involving arithmetic
and geometric means. As regards the task on arithmetic mean, we wanted to find
out whether the respondents would calculate the average from a formula. We are
aware that just because someone did not write down the calculations, it does
not mean that they did not perform them in their memory. On the other hand,
however, any traces of calculations found in the respondents written answers would
be a clear evidence of a procedural approach. In the case of the second task, we
wanted to check whether the respondents recognize in it, a situation in which the
geometric mean, not the arithmetic one, should be calculated. This type of task
appears in a popular secondary school mathematics textbook, so teachers working
at this level of education are likely to be confronted with this topic in their near
future.

3.2. Participants
The study involved 59 students entering undergraduate programs in mathe-
matics (hereafter referred to as ‘students’) and 14 teachers (hereafter called ‘teach-
ers’) who entered postgraduate programs qualifying them to teach mathematics as
another subject. Among the students, 8 people had taken the high school leaving
exam only at the basic level, 48 people at the basic and extended level, and 3
people did not provide information on this matter.
Among the teachers surveyed, 1 person did not complete a short questionnaire
asking for information about age, former education and previous work experience.
Of the remaining 13 persons, there were 3 who indicated education in early child-
hood and pre-school education, another 3 in accounting and finance, and also 3 in
philology (German or Polish). Each of the following disciplines was mentioned by
two teachers: management, physics, chemistry or computer science. Single persons
also indicated theology, logopedics, physical education and sociotherapy.
Three persons declared that they had no work experience as teachers. Two
others had only experience of working in a day-care center, kindergarten or nursery,
or as a supporting teacher or physical education teacher. One person declared
to teach physics at a university, another to give private lessons in mathematics,
including preparation for the international high school final exams. One person,
[16] Barbara Barańska, Małgorzata Zambrowska

who had completed a PhD in chemistry, had been teaching physics and computer
science at secondary school in addition to chemistry. The others indicated that they
were employed in primary or secondary schools accordingly to their educational
background.
What the students and teachers have in common is that both groups began
their studies to obtain preparation for teaching mathematics. An important dif-
ference lies in the fact that while the students have 5 years to acquire relevant
knowledge and skills, the teachers are expected to achieve the same competencies
in three semesters, i.e. 1.5 year. The formal qualifications that both groups are
going to obtain at the end will be the same.

3.3. Organization of the study

A group of students were recruited at the university where one of the authors
works. Undoubtedly, the new situation and the feeling of obligation influenced
strong commitment in completing the tasks and reliably handing over to the test
organizers all the notes taken while solving the tasks.
The students were informed that the test was only for diagnostic purposes and
was not subject to assessment. Students were asked to solve the tasks on the day
the academic year began, when they had not yet taken any classes. When solving
the tasks, they could only rely on the knowledge they had acquired in their previous
education. The study conducted on this group was a pilot study and was aimed
at final verification of the tasks to be used in the study of the second group. The
tasks were carefully analyzed and selected beforehand, and the test sets received
positive reviews from four experts in mathematics education. The pilot study did
not reveal any problems with understanding the content of instructions, and it did
not result in any changes to the test. Therefore, the results obtained during this
study provide a certain point of reference for subsequent groups tested with the
same set of tasks.
Due to pandemic restrictions, students were divided into 6 groups which at-
tended the test in a hybrid mode. Three consecutive blocks of meetings were
scheduled on the day of the study. Within each block, the tasks were solved by
two groups, with one group solving the tasks in a room at the university building
and the other group completing the test remotely at the same time. A total of 31
people completed the tasks stationary, and 28 remotely. Those who took the test
remotely completed the closed and gap tasks via Google Form and were asked to
send photos or scans of their solutions to these tasks together with solutions to
open tasks that were also given to the students.
Because of the pandemic restrictions, far less teachers participated in the study
than was originally expected. At the planning stage of the study, face-to-face meet-
ings with teachers were intended to take place during assemblies at the universities
where the studies are conducted. Due to the pandemic, the stationary classes were
replaced by remote meetings. In order to reach the teachers, we attempted to
contact first the different universities and centers conducting postgraduate studies
qualifying for teaching mathematics as a second subject. Some of the centers did
not respond to our requests, some stated that they did not currently offer the
courses we were interested in, some declined to help us contact students and a few
What do we mean by mean? [17]

centers promised to forward the message we had prepared for their students to in-
vite them to participate in the study. However, even when the invitation reached
a particular teacher, he or she could make the decision voluntarily. We suspect
that many teachers did not receive our message, and of those who did, a signifi-
cant number were unwilling to participate. Ultimately, the results we were able to
collect were obtained in two periods XI-XII 2020 and X-XI 2021.
For the convenience of teachers solving the test remotely, the tasks they were
asked to solve were divided into two blocks: a block of closed tasks, gap tasks and
true-false tasks, which could be answered via Google Forms, and a separate block
of open tasks, where the solution had to be written down in full on a piece of
paper and sent in the form of a scan or a photo. Students had both of these blocks
combined into one form.

3.4. The tasks

In the test, which was completed by both students and teachers, there were
a total of 45 tasks, including 6 true-false tasks, 30 closed or gap tasks and 9 open
tasks. In addition, a group of teachers received 5 open tasks related to the teaching
competences at a separate time, which is not the subject of our attention here. In
this article, we report the results obtained in the two tasks on the averages that
were included in the test. Both tasks were the gap tasks.
Task 1
The content of the first task we would like to present here was as follows:
The arithmetic mean of the numbers x and y is equal to 6. The arithmetic
mean of the numbers x + 10 and y + 10 is equal to ...
This task is similar to some of the examples from the 7th grade textbook that
we referred to earlier in the text. If the arithmetic mean of two numbers x and
y equals 6, it means that each of these numbers could be replaced by 6 and the
sum of the two numbers would remain the same. If we increase each number by
10, it means that now each of the newly obtained number could be replaced by
16, which is the arithmetic mean to be found. No calculation is needed for solving
this task.
Task 2
The second task was the following:
In a newly built housing complex, the number of inhabitants increased in two
consecutive years by 69% and 21% respectively. The average annual percentage
increase in the number of residents of the estate was .... %
The task is similar to one of the examples presented in the high school math-
ematics textbook. By asking about the average annual percentage increase we are
asking what the percentage increase should be if we wanted it to be equal in each
year and leading to the same result as the two changes mentioned in the task. This
however, may not be so obvious to the students, as it is even nowhere explained
in the textbook in such a manner. By asking respondents to solve such a task, we
wanted to observe whether they would determine the geometric mean or interpret
the task as a question about the arithmetic mean of the percentage changes. As
the formulation of the task itself may be ambiguous for the recipients (however, we
[18] Barbara Barańska, Małgorzata Zambrowska

decided to keep the wording close to the one from the textbook), we were curious
if anyone would comment on it in their answer.
It is worth mentioning here that the solution involving the geometric mean does
not require the use of a calculator if one performs the calculations cleverly, which
also proves one’s skills, a certain mathematical proficiency, as well as a developed
number sense. Denoting by a the initial number of inhabitants of the housing
estate, one should find such a number p, that:

p2 · a = 1, 21 · 1, 69a.

Hence:
p2 = 1, 12 · 1, 32
2
p2 = (1, 1 · 1, 3)
p2 = 1, 432
p = 1, 43.
Thus the answer is 43%.

3.5. Results
Task 1
The results obtained in Task 1 are shown in the table below:

Students (N=59) Teachers (N=14)

Answer Number of answers Answer Number of answers
16 46 16 12
8 4 60 1
11 2 16 and 36 1
Other 4
No answer 3

Students’ solutions
We found written records of the solutions to this task in the papers of 25 out
of 31 students solving the tasks stationary and in 19 out of 28 scans submitted
by the students writing the test remotely. This means that 44 of the 59 students
writing the test needed a record of the performed calculation. These students did
not appeal to a conceptual understanding of the arithmetic mean, but performed
calculation using an algorithm learned back in primary school. This observation
alone shows that the arithmetic mean, which is perhaps the mean most familiar
to the students, is treated by the respondents on a largely procedural basis. As
for the other 15 students, we do not know how they proceeded. The calculations,
even if done in this task, are not tedious and they could easily be done in mind.
It came as no surprise to us that most students answered the question posed
in the task correctly. The most frequent solution was the following:
Task 1, example 1
x+y
=6
2
What do we mean by mean? [19]

x + y = 12
x + 10 + y + 10 12 + 20
= = 16
2 2
Two students solved the task without taking this second step:
Task 1, example 2
x+y
=6
2
x + 10 + y + 10 x + y 20
= + = 6 + 10 = 16
2 2 2
In several works, the arithmetic mean was determined as follows:
Task 1, example 3
x+y
=6
2
x + y = 12
x = 12 − y (or y = 12 − x)
x + 10 + y + 10 12 − y + 10 + y + 10 32
= = = 16
2 2 2
Four students obtained number 8 as their final result. We have a record of such
a solution from three of them. In each case, the error was the same: the student
treated x + 10 + y + 10 as the sum of four numbers and performed the following
division:
Task 1, example 4
x + 10 + y + 10
= 8.
4
It is likely that the answers indicating number 11 resulted from calculating the
sum of 6 (half of the sum of x and y) and 5 (half of the number 10), but we did
not find a record of such a solution in any paper with this result.
The other answers given by the students were: 17, 2, 22 and 5,5. The result 17
was obtained because 32 was incorrectly divided by 2 (the method was valid, only
there was a calculation error at the end). The result 22 was obtained in this way:
Task 1, example 5
x+y
=6
2
x + y = 12
x + 10 + y + 10
2
12 + 10
The author of this solution either assumed by mistake that x+y 2 = 12 or she
x+10+y+10
mistakenly simplified the arithmetic part of the expression 2 .
The outcome of 5,5 was obtained due to a calculation error combined with
a wrong method adapted by one student:
Task 1, example 6
x + 10 + y + 10 12 + 10
= = 5, 5.
4 4
[20] Barbara Barańska, Małgorzata Zambrowska

We do not know where the result 2 comes from. It appears in the work of a student,
who filled in the form, but did not send a scan of his written solution to this task
(maybe he did not write anything down, or maybe he did not make a scan of
a sheet of paper with the solution to this task). It is possible that the student
wanted to give a different result, e.g. 22, but entered the data incorrectly in the
form. It is difficult for us to comment on this.
Teachers’ solutions
It is also difficult to analyse the results submitted by the teachers. Although
they were asked to send scans showing everything they wrote or drew while solving
the closed and gap tasks, we received such a scan showing the calculations made
in the task in question only from one teacher. His solution was the same as given
in example 1. We can only assume that other correct answers in this group were
also obtained via methods similar to those used by the students.

Task 2

Students (N=59) Teachers (N=14)

Answer Number of answers Answer Number of answers
No answer 28 No answer 2
45% 20 45% 6
43% 1 52% 3
Other 10 43% 1
Other 2

Among the “other” results, the following appeared:

• In the group of students: two people indicated a result 52.245%, of the re-
maining results, each was indicated by one person: 35,49%; 48,5%; 58%; 44%;
40%; 52%; 50% and 22%;

• In the group of teachers: one person wrote 52,245%, the other wrote 12%.

Students’ solutions
In this task, any written responses appeared in the works of 19 out of 31
students writing the test stationary (with two works containing only the numbers
given in the task), and in the works of 8 students working remotely.
Eight students’ solutions included the calculation of the arithmetic mean. In
six of them, the calculation was as follows (with or without the percentage sign):
Task 2, example 1
69% + 21%
= 45%,
2
and two people, except of using the wrong method, also made another error when
inserting the numbers:
Task 2, example 2

69% + 28% 97%

= = 48, 5%
2 2
What do we mean by mean? [21]

or in calculations:
Task 2, example 3
69 + 21 80
= = 40.
2 2
There were six students who, despite starting with calculations, most of which
were very tedious, eventually gave an answer of 45%, most likely arriving at this
result by ultimately abandoning complicated counting and finding the arithmetic
mean in memory. It can be assumed that the ”ugly” calculations led students to
consider them impossible and wrong. Here are two examples of the situation just
described:
Task 2, example 4

Task 2, example 5
169% − I

1, 21 · 169% ≈ 204, 5%

204, 5 + 169 = 373, 5

373, 5 : 2 =
In both examples it can be seen that the intention of the initial calculations was
good. In Example 4, the student wanted to calculate 21% of the number 1,69
and then add the result to that number. In Example 5, the student immediately
calculated 121% of 169%. While in the case of the author of the solution in example
4 we do not know what he intended to do next, example 5 shows not only an
unjustified addition, but also a plan to calculate the arithmetic mean at the end.
Four students started with some calculations, but did not come up with any
answer. This was the case, for example, in the works of students whose solutions
were the following:
Task 2, example 6
[22] Barbara Barańska, Małgorzata Zambrowska

The calculations done by the student in Example 6 were as follows:

1, 69 · 1, 21 = 2, 0449

0, 21 · 0, 69 = 0, 1449
0, 69 + 0, 1449 = 0, 8349
Finally, the student wanted to determine the arithmetic mean of 69% and 21% of
69%:
0, 69 + 0, 21 · 0, 69
.
2
Perhaps what the student was trying to do, but got lost during partial computa-
tions, was what could be done if we knew that a certain number x was increased
by 69% and then by 21% and the question was how much (by what value, not by
what percentage) the number increased on average.
x – initial value of a number
1, 69x – value of the number after the first change;
0, 69x – the rate of first increase
1, 21 · 1, 69x – value of the number after the first change;
0, 21 · 1, 69x – the rate of second increase
Now, the arithmetic mean
0, 69x + 0, 21 · 1, 69x
2
would inform us what the average increase was if we were interested in the size
(value) of the increment and not the percentage change in the value of the number.
For instance, if our initial number was 40, and the number would be increased
by 50%, this would result in 60. We would then get 90 by increasing 60 again
by 50%. We could say that the average change in the value of the number was
(20 + 30) : 2 = 25. Perhaps the reasoning of the author of example 6 was similar
to the above. This would explain how the student interpreted the question posed
in the task.
Task 2, example 7

The author of Example 7 in her written solution was also calculating the arith-
metic mean. What was calculated was the average of the number of inhabitants
What do we mean by mean? [23]

after the first and after the second year. The result obtained by the student would
indicate that on average in each of these two years, about 86% more people lived
in the estate than at the beginning.
The following solution has ultimately led the student to give 35,49% as an
answer:
Task 2, example 8
0, 21 · 1, 69 = 0, 3549
1, 69 + 0, 3549
The result of 35,49% actually says that the increase of 21% in the following year
compared to the previous year was 35,49% of the initial population.
A final answer of 44% came from a student who in his work also got the number
0,3549:
Task 2, example 9

A student whose solution is shown below, wrote 58% as an answer:

Task 2, example 10

x1 = year 1 x1 = 69%y + y

((69%y + y) + 21%y)
x2 = year 2 x2 =
2
A result of 52,245% was given by two students. In both cases this was due to the
following calculation:
Task 2, example 11
1, 69 · 1, 21 = 2, 0449
2, 0449 − 1 = 1, 0449
1, 0449 : 2 = 0, 522245
One person gave a result of 50% and this was achieved as follows:
Task 2, example 12
1−x
2 − 1, 69x
3 ∼ 2x
4, 69x
≈ 1, 5x.
3
[24] Barbara Barańska, Małgorzata Zambrowska

A good answer was given only by one student. In the scan she sent, we can see the
following written entry:
Task 2, example 13
x → 1, 69x → 1, 69 · 1, 21x = 2, 0449x
We can assume that the student calculated the root of 2,0449, obtained 1,43, and
interpreted this result as an average annual percentage increase of 43%.
The lack of a written solution for this task could indicate either that a student
had no clue how to tackle the task or that (s)he believed the task was trivial and
all that needed to be done was to find the arithmetic mean of two numbers, which
required only simple mental arithmetic. The lack of a written solution co-occurred
in the group of students with the lack of an answer as many as 22 times, while the
lack of a written solution and an answer of 45% occurred 11 times. This shows that
there were twice as many students who were helpless with this task as those who
were confident that the result was an arithmetic mean which could be calculated
in memory. None of the students expressed doubts about the wording of the task,
but it is possible that some of those who did not give an answer were confused
about what mean was meant.
Teachers’ solutions
Only one of the teachers sent a scan of a written solution to this task. It was
the teacher who gave a correct answer. His solution was the following:
Task 2, example 14

The teacher was aware that the mean being sought was the geometric mean.
His result is correct, only the calculations may not have been so complicated.
In the case of 6 teachers who gave 45% as an answer we may assume that the
arithmetic mean was calculated in memory. We also suspect that the result of 52%
was obtained by rounding up the result of 52,245%, that was given by two students
and another teacher. It is possible that the solutions of these teachers were similar
to the one given in example 11. It is difficult for us to justify the result of 12% given
by one of the teachers. We also do not know whether the teachers who did not
answer to this task wrote down any calculations. We hypothesize that the teachers
were reluctant to share their notes with us, which they may have considered to be
a rough draft, not suitable for showing to anyone.

4. Limitations of the study

Before we move on to summarize the findings, we would like to point out that
the study we conducted had some limitations that need to be taken into account
What do we mean by mean? [25]

when drawing conclusions. The number of participants in the study was not big
enough to give us the basis for generalizing the results to the whole population
of students or teachers undertaking postgraduate courses qualifying for teaching
mathematics. The under-representation of the latter group in the study was due to
the pandemic driven measures strongly affecting the conduct of the study. Espe-
cially the restrictions imposed by the Polish government have led to changing the
form of teaching on postgraduate programs from stationary to remote. The lack of
opportunity for direct, face-to-face contact with the study participants precluded
additional, follow-up questions that would have helped to establish, for example,
how the solvers who wrote only their answer thought about the task when taking
the test.
However, the revealed difficulties of the respondents should be taken into ac-
count in the training of future teachers. The results we obtained allow us to state
with certainty that in the group of pre-service teachers there are persons who may
have problems in solving some tasks concerning means. Moreover, it should not
be presumed that, due to many years of accumulating mathematical experience at
different stages of mathematical education, candidates for the teaching profession
have developed conceptual understanding of means.

5. Conclusions and recommendations

As the organizers of the study, we were interested in not only knowing the
results obtained by prospective mathematics teachers, but also in seeing how they
arrived at these results. However, it is surprising that at least 45 of 73 respon-
dents needed to write down the solution to the first task. This means that, for
these people, handling the arithmetic mean is strictly related to performing some
calculations and applying a well-known algorithm.
It is even more disturbing that as many as 30 out of 73 respondents from
both groups did not give any answer in task two, and only 2 respondents were
able to see the geometric mean there and give a correct answer. Our opinion is
that the geometric mean is given too little attention at school, and its inclusion
in the textbook, which has been presented, does not sufficiently explain when
this mean should (could) be used and why. The geometric mean seems to be not
intuitive enough. The current way of introducing this mean does not appeal to
the imagination of the students, does not allow them to get a sense of what the
nature of this mean is and what role it plays for a given set of numbers. Special
attention may also need to be paid to the wording of the tasks themselves, so that
the problem for the learners is not to figure out what they are actually being asked
for, but to make an intellectual effort to solve the problem.
In the light of the obtained results, we strongly suggest that during the classes
on didactics of mathematics a special attention be paid to the tasks 4, 7 and 13 from
pages 300-301 of the 7th grade textbook referred to in this paper. Mathematics
educators should also encourage prospective teachers to formulate analogous tasks
concerning the geometric and harmonic mean, and thus investigate their nature.
One might also be tempted to explore the relationship between the means and
other concepts, e.g., it is relatively easy to show (and may be a big surprise to
[26] Barbara Barańska, Małgorzata Zambrowska

prospective teachers) that the logarithm of the geometric mean of a set of positive
numbers is the arithmetic mean of the logarithms of the individual numbers.
Lastly, we want to suggest a way of introducing arithmetic, geometric and
harmonic means that makes it possible to think and talk about them in a unified
way.
Arithmetic mean
The arithmetic mean answers the question: if, in a given set of numbers, all
the numbers had the same value, what would that value have to be for the sum of
all the numbers to be the same?
For example: for the numbers 3 and 5, the arithmetic mean is 4, because if we
replace each of the numbers 3 and 5 with 4, the sum of the two numbers remains
the same: 3 + 5 = 4 + 4.
Method of determination: (for a set of n numbers) we add all the n numbers
in the set and then divide the resulting sum by n, the number of components.
Geometric mean
The geometric mean answers the question: if, in a given set of n positive
numbers, all the numbers had the same value, what would that value have to be
in order for the product of all the numbers to be the same?
For example: for the numbers 2 and 8, the geometric mean is 4, because if we
replace each of the numbers 2 and 8 with 4, the product of the two numbers is the
same: 2 · 8 = 4 · 4.
Method of determination: (for a set of n positive numbers) multiply all n
numbers from the set, then calculate the nth degree root of the resulting product.
Harmonic mean
The harmonic mean answers the question: if the reciprocals of each positive
number in a given set were replaced by the reciprocal of one, the same number,
what would that number have to be to give the same sum of reciprocals?
For example: for the numbers 3 and 5, the harmonic mean is 154 , because if we
4
replace the inverse of each of the numbers 3 and 5 with 15 , the sum of the two
inverses in each case will be the same: 13 + 15 = 5+3 8 4 4
15 = 15 = 15 + 15 . So, for the
positive numbers a and b, we seek to find such a positive number c that:
1 1 1 1
+ = +
a b c c
Method of determination: (for a set of n positive numbers) the number n is divided
by the sum of the inverses of each number.

Acknowledgement:
This work has been supported by the National Science Centre, Poland (grant
number: 2018/31/N/HS6/03976).

References
Bakker, A. (2003). The early history of average values and implications for edu-
cation. Journal of Statistics Education, 11 (1).
What do we mean by mean? [27]

Bakker, A., and Gravemeijer, K. P. (2006). An historical phenomenology of mean

and median. Educational Studies in Mathematics, 62 (2), 149-168.
Batanero, C., Godino, J. D., Vallecillos, A., Green, D. E., and Holmes, P. (1994).
Errors and difficulties in understanding elementary statistical concepts. In-
ternational Journal of Mathematical Education in Science and Technology,
25 (4), 527-547.
Cai, J. (1995). Beyond the computational algorithm: Students’ understanding of
the arithmetic average concept. In M. Luciano, D., Carraher (Eds). Proceed-
ings of the 19th International Conference for the Psychology of Mathematics
Education (PME) (Vol. 3, 144-151).
Cewe, A., Nahorska, H., Pancer, I. (2003). Tablice matematyczne. Wydawnictwo
Podkowa.
Chronowski, A. (1999). Podstawy arytmetyki szkolnej. Liczby wymierne, rzeczy-
wiste i zespolone, cz. 2, Wydawnictwo Kleks, Bielsko-Biała.
Guimarães, G., Gitirana, V., Marques, M., and dos Anjos, D. (2010). The con-
cept of mean by primary school students. In C. Reading (Ed.), Data and
context in statistics education: Towards an evidence-based society. Proceed-
ings of the Eighth International Conference on Teaching Statistics. Voor-
burg, The Netherlands: International Statistical Institute. Retrieved from
http://icots.info/icots/8/cd/home.html
Gfeller, M. K., Niess, M. L., and Lederman, N. G. (1999). Preservice teachers’
use of multiple representations in solving arithmetic mean problems. School
Science and Mathematics, 99 (5), 250-257.
Heath, T. L. (1981). (1921). A History of Greek Mathematics. New York: Dover.
Hess, A. L. (1961). A Geometric Construction for the Arithmetic Mean, the Geo-
metric Mean and the Harmonic Mean of Two Positive Numbers. School Sci-
ence and Mathematics, 61 (1), 45-46.
Jacobbe, T., and Carvalho, C. (2011). Teachers’ understanding of averages. In
Teaching statistics in school mathematics-Challenges for teaching and teacher
education (pp. 199-209). Springer, Dordrecht.
Landtblom, K. K. (2018). Prospective teachers’ conceptions of the concepts mean,
median and mode. In Students’ and Teachers’ Values, Attitudes, Feelings and
Beliefs in Mathematics Classrooms (pp. 43-52). Springer, Cham.
Leavy, A., and O’Loughlin, N. (2006). Preservice teachers understanding of the
mean: Moving beyond the arithmetic average. Journal of mathematics teacher
education, 9 (1), 53-90.
Leon, M. R., Zawojewski, J. (1990). Use of arithmetic mean: An investigation of
four properties issues and preliminary results. In D. Vere-Jones (Ed.), Pro-
ceedings of the Third International Conference on Teaching Statistics: School
[28] Barbara Barańska, Małgorzata Zambrowska

and General Issues, (Vol. 1), International Statistical Institute, Voorburg, pp.
302–306.
Mokros, J., and Russell, S. J. (1995). Children’s concepts of average and represen-
tativeness. Journal for Research in Mathematics Education, 26 (1), 20-39.

Star, J. R. (2005). Reconceptualizing procedural knowledge. Journal for Research

in Mathematics Education, 36 (5), 404-411.
Strauss, S. and. Bichler, E. (1988). The development of children’s concepts of the
arithmetic average. Journal for Research in Mathematics Education, 19 (1),
64-80

Watson, J. M., and Moritz, J. B. (2000). The longitudinal development of under-

standing of average. Mathematical Thinking and Learning, 2 (1-2), 11-50.
School textbooks referred to in the text
Primary school textbooks

Dobrowolska, M., Jucewicz, M., Karpiński, M., Zarzycki, P. Matematyka z plusem

5. Podręcznik. (No. on the list of the Ministry of National Education:
780/2/2018). Gdańskie Wydawnictwo Oświatowe.
Dobrowolska, M., Jucewicz, M., Karpiński, M., Zarzycki, P. Matematyka z plusem
6. Podręcznik. (No. on the list of the Ministry of National Education:
780/3/2019). Gdańskie Wydawnictwo Oświatowe.
Dobrowolska, M. (Ed.), Matematyka z plusem 7. Podręcznik. (No. on the list of
the Ministry of National Education: 780/4/2017). Gdańskie Wydawnictwo
Oświatowe.
Secondary school textbook

Kurczab, M., Kurczab, E., Świda, E. Podręcznik do liceów i techników. Zakres

rozszerzony. Klasa 1. (Podręcznik dla absolwentów szkoły podstawowej. No.
on the list of the Ministry of National Education: 979/1/2019). Oficyna
Edukacyjna Krzysztof Pazdro.

Barbara Barańska
Pedagogical University of Krakow,
Poland
e-mail: barbara.baranska@up.krakow.pl

Małgorzata Zambrowska
The Maria Grzegorzewska University,
Poland
e-mail: mzambrowska@aps.edu.pl