Investigation of

probability distributions
using dice rolling simulation
Stanislav Lukac & Radovan Engel
Pavol Jozef Safarik University, Slovakia
<stanislav.lukac@upjs.sk>

T his paper deals with the evaluation of dice outcomes in two problem-
solving situations. Dice are considered one of the oldest gambling
devices and thus many mathematicians have been interested in various
dice gambling games in the past. We can take the famous mathematicians
as an example: Blaise Pascal and Pierre de Fermat applied the theory of
probability to solve problems connected with dice rolling. Dice have been
used to teach probability ever since.
Dice rolls can be effectively simulated using technology. The National
Council of Teachers of Mathematics (NCTM, 2000) recommends that
teachers use simulations to give students experience with problem situa-
tions that are difficult to create without technology. Computer technologies
furnish visual images of mathematical ideas; they facilitate organising and
analysing data; and they compute efficiently and accurately. Similarly, in
accordance with the Standards for Excellence of The Australian Association
of Mathematics Teachers (AAMT, 2006), a variety of appropriate teaching
strategies is incorporated in the intended learning experiences, enhanced
by available technologies and other resources. Real dice rolling and
recording of outcomes are crucial in an introductory phase of the problem-
solving process. These activities develop students’ abilities to organise data
for effective analysis. Students can draw different conclusions by carrying
out a small number of random experiments. Based on these differences,
students should realise that a large number of random experiments is
necessary to discover general relationships.
A spreadsheet program can be used for simulation of random events
because it enables users to generate random numbers into the selected cells
of a table (e.g., Microsoft Excel offers several tools for generating random
numbers). In this activity we use the basic mathematical function RAND for
the simulation of dice outcomes. Values of this function are random
numbers from the interval 〈0, 1) generated according to a uniform proba-
bility distribution. As we need random natural numbers from 1 to 6, we
multiply the values of the function RAND with the number 6. We get
random numbers from the interval 〈0, 6). To cut off the decimal part of real
numbers, functions INT or TRUNC can be used. By applying one of these
functions to random numbers from the interval 〈0, 6), we obtain integers

30 amt 66 (2) 2010

 from the set {0, 1, 2, 3, 4, 5}. Therefore, it is necessary to add 1 to the partial
result to simulate dice rolling. The final formula looks as follows:

=1 + INT(6*RAND()) (1)

Rolling the number 6

The first problem-solving situation can be presented to students as a game.

One student from a group rolls a die; the other students guess the number
of trials needed to roll the number 6 for the first time. Having repeated the
game at least 10 times, students are asked to answer the following ques-
tion:
Imagine that you have the following two options:
1. The number 6 will be rolled in the second trial.
2. The number 6 will be rolled in the fifth trial.
If you were to bet money, which one would you choose?

Students are required to consider which option is more probable. After

the introductory phase of experimentation in a paper-and-pencil environ-
ment, they then proceed to a computer simulation.
Students create an MS Excel table to model outcomes of a series of dice
rolls. The outcome 6 has to occur exactly once in each series, whereby
rolling the number 6 terminates each series. Frequencies of rolling the
number 6 will be evaluated in the table. In order that the obtained frequen-
cies correspond to theoretical probabilities, it is essential to simulate a large
number of random experiments. This connection between probability and
frequency will be used to evaluate the outcomes of series of dice rolls. A
large number of series of rolls will be modeled in the table, and the
outcomes of each series will be placed into the rows of the table. Since it is
impossible to determine in advance the number of trials necessary to roll
the number 6 for the first time, the table should include generated rows
with a variable number of cells completed by the number 6. Visual Basic for
Applications (VBA) can be used to generate such a table. The basis of the
VBA macro is the function RND where values are repeatedly entered into
cells using cycles. The macro can be executed by means of a button placed
on a sheet. The macro listed below is designed to generate the table with the
first row beginning in cell A1. The table contains the given number of rows

Private Sub CommandButton1_Click()

VBA macro for generating Dim Count As Integer, I As Integer, J As Integer, R As Integer
the table with random Count = InputBox("Specify number of series")
outcomes of dice rolls Range("A1").Select
For I = 1 To Count
J=0
Do
J=J+1
R = Int(6 * Rnd) + 1
ActiveSheet.Cells(I, J).Value = R
Loop Until R = 6
Next I
Range("A1").Select
End Sub

amt 66 (2) 2010 31

 (variable Count) with random outcomes of dice rolls completed by the
number 6.
Another method—based on entering and copying the formulas—can be
used to simulate several series of dice rolls. To make the copying of the
formula easy, we suggest creating a table with a constant number of cells
in rows. Each row will contain 30 random numbers from the set {1, 2, 3, 4,
5, 6} and will represent one series of dice rolls. We suggest that students
generate at least 5000 series of dice rolls. It is necessary to ensure that no
other outcomes of dice rolls will be written into the cells of each row after
rolling the number 6. The number 0 will be entered in these cells and it will
indicate the state after rolling the number 6 in an individual series of rolls.
It may happen that the number 6 will not appear even after 30 rolls.
However, these cases are rare and will not be included in the final evalua-
tion due to counting the number 6 in the table columns.
Having created the table header (see Figure 2) we enter the following
formula into cell B2:

=IF(OR(A2=6,A2=0),0,1 + INT(6*RAND())) (2)

Since a text value is stored in cell A2, the condition in the function IF is not
fulfilled and therefore a random number from the specified set is entered
into cell B2. By copying this formula into a continuous area containing
5000 rows and 30 columns, students can generate outcomes of 5000 series
of dice rolls very quickly. If the number 6 has not been rolled in an indi-
vidual series of dice rolls, a new outcome will be generated, otherwise a
value of 0 will be entered into the cell. Values 0 will repeat until the end of
the series after the first occurrence of value 0. Zero values, gathered in the
generated table, do not represent outcomes of dice rolls and therefore
should not be displayed. Setting of a conditional format in the table can
improve visual orientation in the outcomes. To hide zero values in the table,
the text colour can be set to white by using the <Format> command for each
cell containing value 0 in the area B2:AE5001. Figure 1 shows the dialogue

Figure 1 Conditional formatting for cells containing the zero value.

Figure 2. The part of the table with outcomes of series of dice rolls.

32 amt 66 (2) 2010

window <Conditional Formatting> with the corresponding settings. The left
upper cut-out of the table is shown in Figure 2.
Let Ai be the random event that the number 6 is rolled for the first time
in the i-th roll. We can determine the number of occurrences of the number
6 in each column of the table using the function COUNTIF for comparing
probabilities p(Ai ), i = 1, ..., 30. The bottom part of the table containing
summary results for the first eight rolls in series is shown in Figure 3.
The frequency of rolling the number 6 in the first roll of a simulated
series of dice rolls is calculated in cell B5002, where the following formula
is entered: =COUNTIF(B2:B5001,"=6"). We copy this formula into the adja-
cent cells in the row until cell AE5002. The sum of frequencies in the area
B5002:AE5002 does not need to be equal to 5000, because it may happen
that the number 6 will not appear in some series of dice rolls. The number
of “unsuccessful” series is calculated in cell I5003 according to the formula,
which is shown in the formula bar in Figure 3.

Figure 3. Summary results and the frequency column graph.

The frequencies of rolling the number 6 in the first eight rolls are
displayed in the column graph in Figure 3. We can generate new random
numbers using the F9 key. New dice outcomes enable students to investi-
gate changes in summary results and also in the graph. The histogram
shows that the frequency of rolling the number 6 is the largest in the first
roll and then gradually decreases. This distribution of frequencies can be
justified by the calculation of probabilities:

5(
i −1)
p ( A1 ) = , p ( A2 ) = i , …, p ( Ai ) =
1 5 1 1
i ,i ∈N
6 6 6 6 6

This probability distribution is an example of geometric probability distribution.

Sum of the scores on the dice

The second problem is closely related to the dice game called Craps. Rolls
of two dice and different values of the scores on the dice are the basis of this
game. Considering the fact that the probability of the individual dice roll

amt 66 (2) 2010 33

 outcomes is the same, it might seem that individual sums of the scores are
also equally probable. Even the mathematician Gottfried Wilhelm Leibniz
wrongly supposed that the sum of 11 is as probable as the sum of 12.
However, practical experience differs. Students should decide which sum, 9
or 10, occurs more frequently when rolling the three dice.
Formula (1) can be used to generate dice roll outcomes. To begin, we
create a table with three columns representing 5000 outcomes of rolls of the
three dice. Then we add a new column, Sum, containing the sums of the
scores on the dice. Considering the fact that operations with the data in the
table change the values of the RAND function, it is necessary to transform
generated data to constant values. Having selected the whole table by the
key combination <Ctrl + *>, we copy the table to the clipboard and then we
paste the content of the clipboard as constant values to a new worksheet
using the command <Paste Special>. In order to determine the frequencies
of the individual sums, it is necessary to sort the table by values in the Sum
column using the corresponding icon on the Standard Toolbar, and then
use the command <Subtotal> from the <Data> menu. This command
enables us to determine the numbers of successive equal values after each
change in the Sum column. The dialogue window with the corresponding
settings is displayed in Figure 4.
Figure 5 shows several rows of the ordered table after the application of

Figure 4. The determination of frequency of sums using the command Subtotal.

Figure 5. The part of the table after application of the command Subtotal.

34 amt 66 (2) 2010

 the command <Subtotal>. The buttons 1, 2, and 3 in the left upper corner
change the level of the overview of the calculated data in the table. Figure 5
shows the lowest level of the overview marked with the number 3. Level 2
displays the frequencies of individual sums and the total number of rolls.
This summary table of the frequencies of individual sums (see Figure 6) can
be used to create the corresponding column graph shown in Figure 7.

Figure 6. The frequency Figure 7. The frequency column graph of individual sums.
of individual sums.

The sequence of steps described above (beginning with copying the

random numbers using the command <Paste Special>) can be repeated to
process another series of 5000 rolls of the three dice. The reason is that
random numbers generated by means of the function RAND in the original
table were changed when editing the copied table. Advanced students could
record the above sequence of steps in the macro to automatise repeated dice
rolling simulations.
Finally, the teacher should explain how the experiment results can be
theoretically justified. Galileo Galilei was interested in the sums of the
scores on the dice and explained why the sum 10 occurs more frequently
than the sum 9 when rolling the three dice. The justification is based on the
list of all possibilities of how the sums 9 and 10 can be obtained with the
three dice. There are more possibilities for the sum10 than for the sum 9
and this is the reason why the sum 10 occurs more frequently than the sum
9 in a large number of repeated random experiments. Analogously, median
sums are the most probable when summing the scores on several dice,
because there are more possibilities to obtain them. An increasing number
of dice and an increasing number of repeated random experiments result in
the fact that the distribution of frequencies of the score sums on the dice
approaches to a normal probability distribution.

References
The Australian Association of Mathematics Teachers Inc. (2002, 2006). Standards for
excellence in teaching mathematics in Australian schools. Adelaide: Author.
National Council of Teachers of Mathematics [NCTM] (2000). Principles and standards for
school mathematics. Reston, VA: Author.
Walkenbach, J. (2003). Excel 2003 Bible. John Wiley & Sons.

amt 66 (2) 2010 35