1

Luke Wright
AP Statistics - 7th Period 2014-2015
{{SUBMISSION DATE}}

Can Coffee Make You Smarter?

Abstract: I performed an experiment on students during the school day to measure how coffee
and tea impact their ability to accurately complete math questions as seen on the SAT. I found
that no significant difference existed between the two groups.

2
Table of Contents

1. Cover
2. Table Of Contents
3-5. Report
6.

Works Cited

7.

R-code - T-test

8.

Organized Data

9.

Quiz 1: Initial/Pre-drink

10.

Quiz 2: Post drink

11-12.

Photos of quizzes

3
Report
For many people the first step of their day is a cup of coffee to wake up, and many drink
caffeinated drinks throughout the day to stay alert. There is no doubt that caffeine can help one
stay awake, but I was interested in whether or not it can help a student improve their
performance on difficult problems such as those seen on the SAT. Prior medical studies I found
showed that in children, Caffeine succeeded in improving attentiveness and dexterity, but made
no conclusions about the impact on a students mental performance on challenging problems
(Caffeine Effects). Knowing whether caffeine can improve academic performance is important
information for students, who often drink caffeinated drinks before large exams or to get through
late nights of studying. My hypothesis was that the higher caffeine content of coffee would result
in a great improvement in test scores than tea, which has significantly less caffeine. I
investigated students from Austin High School.
My sample consisted of 30 students from Austin High School, it included both students in
the AGS program and not in it and included varying numbers of all grade levels. One flaw with
my sample is that it was a convenience sample of many friends and classmates, instead of
being a simple random sample. This allowed me to complete my experiment and gather the
needed data, as asking for approximately 20-30 minutes of 30 random students days would
have been highly disruptive to those students and likely resulted in a large number of nonresponses (which would introduce their own issues). However, this also introduced the
possibility of confounding variables and means that my results may be subject to question.
While one could expect the findings at Austin High to be similar to other schools, that would not
be a valid assumption mathematically and thus our findings can only be applied to students at
Austin High.
I performed my experiment by giving participants an initial quiz of 5 SAT math questions,
and then randomly assigning them either coffee or tea, which they were instructed to drink. In

4
order to allow time for the caffeine to enter the participants system, they then waited 10 minutes
before beginning a second quiz of 5 SAT math questions. Participants had as much time as they
needed to complete both quizzes. Both groups were administered identical quizzes and asked
to sign their name. The differences between the number of correct answers on the final quiz and
initial quiz were compared between the two groups and then a two-sample t-test was run to
determine if the differences between the mean differences of the groups were significant. I
found that both groups showed no improvement, on average, from quiz 1 to quiz 2. The coffee
group had a mean difference of exactly zero, whereas the tea group had a mean difference of
-0.1333. However, this alone doesnt conclusively indicate that the caffeine in the coffee had no
impact.
In order to determine if there was or was not a significant impact, I would need to run
two-sample t-test. I chose a two-sample t-test as the two samples are independent of one
another, and we do not know the true population standard deviations, so a t-test is more
appropriate than a z-test. After ensuring that my data was less than 10% of the population I was
interested in, and approximately normal based on the linearity of a normal quantile plot taken
from each sample, I used RStudio to run a t-test. The t-test returned a t-value of just .222, in
order to find my p-value, I looked at the probability that t would be greater than 0.222, which is .
4131. Because our p-value is greater than a reasonable alpha level of .05, we will fail to reject
the null hypothesis that the true mean difference of before and after scores is the same for
coffee and tea.
We dont have evidence that suggests the higher caffeine content of coffee results in any
better performance on the type of mental tasks included in the SAT math section. This was not
what I was initially expecting, but upon further analysis makes more sense. Caffeine is great at
keeping us awake or waking us up in the morning, but a small boost in energy is not a make-orbreak difference between being able to solve a problem midway through the day when most

5
studentss energy levels are already high. The practical side of this research is that I do not think
drinking a large cup of coffee before an exam is likely to result in any better performance, and I
wouldnt recommend that course of action to friends or classmates. It is important to understand
the effects of what we consume, and there are many misconceptions around caffeinated drinks.
Coffee may be a good way to pick you up midway through a long day, but it isnt going to make
you any smarter or somehow allow you to know things you didnt before.

6
Works Cited
Bernstein GA, Carroll ME, Crosby RD, Perwien AR, Go FS, and Benowitz NL. "Caffeine Effects
on Learning, Performance, and Anxiety in Normal School-age Children." J Am Acad
Child Adolesc Psychiatry 33.3 (1994): 407-15. Print.

7
R Code - T-test
> # The difference between final quiz and initial quiz scores of participants
who drank coffee
> coffee <- c(0, -2, 0, 0, 1, 0, -2, 1, -3, 1, 0, 1, 1, 1, 1)
> # The difference between final quiz and initial quiz scores of participants
who drank tea
> tea <- c(1, 3, 2, 1, -2, 2, 1, -2, -3, 0, -2, -3, 0, 1, -1)
> mean(coffee)
[1] 0
> mean(tea)
[1] -0.1333333
>
> # ASSUMPTIONS
>
> # SRS FALSE RESULTS MAY BE SUBJECT TO QUESTION
> # Samples are independent - Another student receiving coffee does not
impact a future trial
> # We know there are greater than 300 students at Austin High
> # check normality of data
> qqnorm(coffee)
> qqline(coffee)
> qqnorm(tea)
> qqline(tea)
> # based on linearity of our NQP, we have approx normal data
>
> # H0: mean(coffee) - mean(tea) = 0
> # HA: mean(coffee) - mean(tea) > 0
>
> # we expect coffee to be greater than tea, data are not paired because
coffee and tea weren't paired, only the differences within each group were
> [Link](coffee, tea, paired=FALSE, alternative="greater")
Welch Two Sample t-test
data: coffee and tea
t = 0.222, df = 24.689, p-value = 0.4131
alternative hypothesis: true difference in means is greater than 0
95 percent confidence interval:
-0.8929514
Inf
sample estimates:
mean of x mean of y
0.0000000 -0.1333333
>
>
> # Because our p-value > .05, we will fail to reject our null hypothesis.
> # We do not have evidence to suggest the caffeine level in coffee
positively affects student performance on math questions compared to tea

8
Data: Final quiz - Initial quiz, organized by experimental group
# Correct on
final quiz - #
correct on initial
quiz
Coffee

Coffee

-2

Coffee

Coffee

Coffee

Coffee

Coffee

-2

Coffee

Coffee

-3

Coffee

Coffee

Coffee

Coffee

Coffee

Coffee

Mean

Tea

Tea

Tea

Tea

Tea

-2

Tea

Tea

Tea

-2

Tea

-3

Tea

Tea

-2

Tea

-3

Tea

Tea

Tea

-1

Mean

-0.133333333333333

9
Quiz 1: Initial/Pre-drink
What is the result when 436921 is rounded to the nearest thousand and then expressed in scientific
notation?
(A) 4.369 times 10^2
(B) 4.369 times 10^4
(C) 4.37 times 10^4
(D) 4.37 times 10^5
(E) 4.4 times 10^5

At the beginning of 2006, both Alan and Dave were taller than Boris, and Boris was taller than Charles.
During the year, Alan grew 2 inches, Boris and Dave each grew 4 inches, and Charles grew 3 inches. Of
the following, which could NOT have been true at the beginning of 2007?
(A) Alan was shorter than Boris.
(B) Alan was shorter than Charles.
(C) Boris was shorter than Dave.
(D) Dave was shorter than Alan.
(E) Dave was shorter than Charles.

A special lottery is to be held to select the student who will live in the only deluxe room in a dormitory.
There are 100 seniors, 150 juniors, and 200 sophomores who applied. Each senior's name is placed in
the lottery 3 times; each junior's name, 2 times; and each sophomore's name, 1 time. What is the
probability that a senior's name will be chosen?
(A) 1 over 8
(B) 2 over 9
(C) 2 over 7
(D) 3 over 8
(E) 1 over 2

In the x y-coordinate plane above, line l contains the points (0, 0) and (1, 2). If line m (not
shown) contains the point (0, 0) and is perpendicular to l, what is an equation of m?
(A) y = -(1/2)x
(B) y = -(1/2)x + 1
(C) y = -x
(D) y = -x + 2
(E) y = -2 * x

If k is divisible by 2, 3, and 15, which of the following is also divisible by these numbers?
(A) k + 5
(B) k + 15
(C) k + 20
(D) k + 30
(E) k + 45

10
Quiz 2: Post Drink

In a survey, a group of students from Westville High School were asked about their favorite movie genre.
Each student in the group selected exactly one movie genre, and the data collected are summarized in
the circle graph above. If 40 more students chose Action than Drama, how many students were surveyed
in total?
(A) 100
(B) 150
(C) 200
(D) 250
(E) 300
If
(A) -4
(B) -3
(C) -1
(D) 2
(E) 3

, for which of the following values of x is y NOT defined?

Which of the following statements must be true of the lengths of the segments on line m above?
1. AB + CD = AD
2. AB + BC = AD - CD
3. AC - AB = AD - CD
(A) 1 only
(B) 2 only
(C) 3 only
(D) 1 and 2 only
(E) 1, 2, and 3

and x=12, then x-y=

11
Photos of Completed Quizzes

12