Mathematical Studies Internal Assessment

An investigation into the probability of winning Rock-Paper-Scissors

Table of Contents

Introduction………………………………………………………………………………….....2
Mathematical processes……………………………………………………………………...4

Discussion……………………………………………………………………………………..7

Validity………………………………………………………………………………………….9

Conclusion…………………………………………………………………………………...10

INTRODUCTION

A common game used to settle bets, arguments or play for entertainment is called

Rock-Paper-Scissors. To play the game, two people count to three and either present

1
with their hand a fist, representing “rock”, a flat hand, representing “paper”, or two

fingers in a V-shape, representing “scissors”. In theory, there is a one in three chance of

winning, having a tie, or losing. However, if you study this further, you can use simple

concepts of human psychology to have an insight as to the likelihood of the next move.

In past research on this subject, it has been predicted that

when a person loses a round, his/her next move will most

likely be “rock”. The winner of the round most often picks “scissors” as the next move,

which makes “paper” the least common pick in a battle of best two out of three.

In this study, I am going to calculate the probability that the choice of play, whether one

choses to throw a “rock”, “paper” or “scissors”, is based on one’s status in the game. In

other words, I will investigate how whether an individual is winning or not winning Rock-

Paper-Scissors affects the next move.

I was curious to see if the majority of participants played their next move based off of

whether they won or lost the previous round, or if each move was thrown at random. To

test this, I sampled eight pairs of participants and ran approximately sixty rounds of the

game. I ran a chi-squared test and used pie charts to record the results. The chi-

squared test ultimately showed whether or not a person’s choice of play is independent

2
of their status of the game, whether a person’s choice is based on winning or not

winning. The pie charts were used to record what play was most often used based upon

the participant’s game status. I choose to use pie charts because they are visually

appealing and can easily show comparisons in percentages out of 100.

To demonstrate a less specific range of data, a chart of overall percentages is

presented. This chart displays overall totals as well as totals by gender. This chart does

not reflect the status of the game. This data was taken and recorded for purpose of

comparison. For example, what percentage of people chose “rock” when they were

winning the game versus how many “rocks” were thrown overall when not considering

status of the game. The purpose of running these tests was to see if the probability of

each play being thrown a third of the time is accurate in a real life situation. There are

two main questions that I attempted to answer throughout the the course of these three

mathematical processes:

1. Is there a correlation between an individual’s choice of play and status in the

game?
2. How does the overall percentage of either “rock”, “paper” or “scissors” compare

to the percentages based on an individual’s status in the game?

Mathematical Processes

To start my data collection, I asked students in my math class to play five rounds of

Rock-Paper-Scissors. I took all of the plays of the participants and calculated the

percentages of each play thrown overall. This data did not account for whether the

players were winning or not winning. The results of the collected data is shown below.

3
 ROCK PAPER SCISSORS

OVERALL 36.8% 32.9% 30.3%

MALE 32.2% 41.9% 25.8%

FEMALE 40.0% 26.7% 33.3%

I also utilized my fellow math students to test my theory that choice of play and status in

the game are not independent. The table below shows the number of chosen plays

based on an individual’s status in the game.

ROCK PAPER SCISSORS

WINNER 6 8 6

NOT WINNER 16 12 12
I chose to run a chi-squared test to analyze this data, in order to test my theory that an

individual’s choice of play and status in the game are not independent of one another.

Chi-squared tests determine how likely the data is due to chance. It measures whether

the observed data and expected values are independent of one another. This required

using the data collected to find the expected value of choice of play. To find the

expected values, I took the total number of one choice of play, for example, 22 “rocks”

and divided by the total sample size, 60. The outcome was then multiplied by the total

number of one game wins which was 20, divided by the total sample size, 60, multiplied

by the total sample size, 60. The results demonstrate that my data,

6 rocks thrown when winning, was lower than what was expected for my sample size.

Below are the expected values after the above process was run for every category.

4
 ROCK PAPER SCISSORS

WINNER 7.33 6.67 6.00

NOT WINNER 14.67 13.33 12.00

After collecting all the necessary data, I ran a chi-squared test. The null hypothesis is

that choice of play and status in the game are independent. The alternative hypothesis

is that the choice of play and status in the game are dependant upon one another. The

formula for this test and the collected data is below:

My calculations are as follows:

To find the degrees of freedom, I took the number of rows minus one multiplied by the

number of columns minus one. There are two degrees of freedom for my test. I chose to

pick a probability value of 0.05, which is the most common throughout various tests in

areas of math and science. This means that the allowed error is 5%. The chart below

will show what the chi-squared value should have been for my degrees of freedom and

chosen probability value.

My chi-squared value should have been approximately 5.99. Since my value was 1.27

which was smaller than the accepted value of 5.99, I could not reject the null

5
hypothesis. With this information, I concluded that choice of play and an individual’s

status in the game are independent of one another.

As my third mathematical process, I chose to make a pie chart of percentages from the

collection of data. To do this, I took the number of each play and divided it by the total

number of either winners or non-winners, depending on the section of relevant data.

This process demonstrates the most frequent choice of play in regards to an individual’s

status in the game. After this mathematical process was complete, I could then compare

the pie charts to the table presented initially, of the overall throws during the experiment.

6
Discussion

7
As seen in the charts above, “paper” was the most common choice of play when an

individual won the round. “Rock” was the most common choice of play if the player

either tied or lost the previous round. Also reflected in the pie charts was that each

move; “rock”, “paper” or “scissors”, was thrown about a third of the time. This

information parallels the table shown initially where each of the plays were also thrown

about a third of the time, regardless of an individual’s status in the game.

Through many calculations, I have concluded that in the game Rock-Paper-Scissors

what someone throws as their next move is independent of whether or not they are

winning. The probability of someone choosing to play a “rock”, “paper” or “scissors”

was close to one-third when considering or not considering one’s status in the game.

This is depicted in both the pie charts and in the table of overall percentages.

I was surprised by the results of my investigation. In past observations, “rock” seemed

to be the most common choice of play, as it seems to be thought of as the most

powerful. Therefore, I predicted that it would be used the most when someone is losing.

I predicted that there would be a large difference between the amount of “rocks” thrown

than the other two. I also predicted that not as many people would chose to play

“paper”, since it seemingly is the weakest form of defense. However, as the results

show, there were more people who threw “paper” when they were winning than either of

the other two options. As I predicted, the most common play was “rock” overall, but

there was not a significant difference between that choice and the choice of “paper” or

“scissors”.

8
The overall percentage of play of “rock”, “paper” or “scissors” compared to the

percentage of the three individual plays determined by status in the game have a

correlation. Each play is thrown about a third of the time. The overall percentages have

a stronger one to three ratio than those based on whether an individual is winning or not

winning.

Validity

Initially, in my chi-squared test, I had expected values below five, which means the data

was too simple, invalid and could not be used for further replication. Of the different

ways I could have approached this situation, I chose to run more trials in order to

achieve values above five. According to the chi-squared test requirements, values

below five are not valid or reliable. In many cases, if the expected value is too low in a

column, you could combine two or more columns to obtain a larger sample in a

particular range. Since my test has variables that were not numerical, I could not

combine rows or columns to increase the value.

I checked my results with the TI-Nspire to make sure my calculations were accurate for

both the chi-squared test and the percentages given. I rounded some numbers to make

the values in the tenth or hundredths place if they extended beyond. Since these values

9
were rounded, they are not precise. However, since this experiment was looking for a

general one to three ratio, the rounding did not affect the validity of the results.

In conclusion, these results can be determined to be valid and could be used for further

research. All chi-squared values were above five and the calculated percentages were

rounded appropriately with respect to the goal of the experiment.

Conclusion

This investigation demonstrates that in the game Rock-Paper-Scissor, a participant’s

status in the current game and their next choice of play are independent of one another.

Each choice of play was thrown overall about one-third of the time. I used the chi-

squared test and percentages through pie graphs and tables to come to these two

conclusions. All math processes were calculated by hand or using simple processes on

a calculator, leading to little or no limitations throughout the investigation. However, I did

use prior math knowledge to run the tests. For example, in the chi-squared test, I

needed to know how to calculate the expected values and the procedure of running the

test itself prior to starting the interpretation of the results. The largest obstacle was

obtaining values above five, since initially I did not have a large enough sample size to

allow my values to be valid.

10
11