The Unofficial Coke vs. Pepsi Challenge Website

The Actual Study

 

I obtained the samples from the master list of all the students.  I took the list, and through Minitab, separated the list into four lists of students from each grade.  From those lists, I took a stratified random sample of 30 students from each grade.  I used Minitab to get a list of 30 students from each grade.  I obtained the list by taking a sample from the columns. The sample that I chose was 30 students from each grade.  I wasn’t really monitoring the amount of males and amount of females that the study had, but it was pretty even.  There were 62 total males and 58 total females.  There were 60 total underclassmen (30 freshmen & 30 sophomores), and there were 60 total upperclassmen (30 juniors and 30 seniors).  There were 18 males and 12 females for the freshmen, 15 males and 15 females for sophomores, 7 males and 23 females for juniors, and 18 males and 12 females for seniors.  I am going by grade and not by grade and gender due to the skewed results.  The periods the students were in was somewhat normally distributed since 60% of the students (72/120) were in 4/5, 6/7, or 8/9.  24 out of 120 students were in each one of those periods.  45 out of the other 48 students were in 1, 2, 3, 10, or 11 and three (all senior males) were disqualified because they went to full-time Polaris.  There are a total of 117 students in the sample pool.  Of these 117 students, 95 responded.  32 students preferred Pepsi, and 63 students preferred Coke.  A copy of the survey is below.

 

Gender:    Male    Female

Grade:    9    10    11    12

Which do you prefer?    Coke    Pepsi

 

Gender:    Male    Female

Grade:    9    10    11    12

Which do you prefer?    Pepsi    Coke

 

I alternated Coke and Pepsi to prevent selection bias.

 

Click here for Graphs

 

Click Here for Raw Data

 

I ran several significance tests from the data.  The first was a two proportion z-test to see if the proportion of females that prefer Pepsi was different than the proportion of males that preferred Pepsi.  I ran the test and got a p-value (p1 - p2)of .949.  The second test was a chi-squared test of the observed and expected values of the proportions of people in each grade that preferred Pepsi were all equal.  The test results for that chi-square test came out with a p-value of 0.626.  The third test was a 1-proportion z-test to see if the sample proportion of the people that prefer Pepsi is equal or not equal to the proportion of people that prefer Coke.  I got a p-value of .004, so I can reject the null hypothesis and say that the proportion of people that prefer Coke is indeed greater than the amount of people that prefer Pepsi.

 

µ1: The true proportion of people that prefer drinking Pepsi

µ2: The true proportion of people that prefer drinking Coke

H0: µ1 - µ2 = 0

Ha: µ1 - µ2 ≠ 0

α = .05

Assumptions:  n = 30, so the sample is so normal

t =

Conclusion: We fail to reject the null hypothesis at the .05 level because our p-value > α.  Therefore, there is enough significant evidence to prove that the true mean proportion of females that prefer Pepsi is equal to the amount of males that prefer Pepsi.

 

µ1: The observed proportion of people that prefer Pepsi Cola.

µ2: The expected proportion of people that prefer Coca-Cola.

H0: µ1 =  µ2

Ha: µ1 ≠  µ2

α = .05

x² = (observed - expected)² / expected

Assumptions:  n = 30, the sample is random.

Category:    Observed:    Proportion:     Expected:     to Chi-Sq

Freshmen         9                  0.25               8                 0.125

Sophomores     8                  0.25               8                 0.000

Juniors             10                 0.25               8                 0.500

Seniors             5                  0.25               8                  1.125

Conclusion: We fail to reject the null hypothesis at the .05 level because our p-value > α.  Therefore, there is enough significant evidence to prove that the distribution of those who prefer Pepsi is any different by grade.

 

µ1: The true proportion of people that prefer drinking Pepsi

µ2: The true proportion of people that prefer drinking Coke

H0: µ1 = µ2

Ha: µ1 ≠ µ2

α = .05

Assumptions:  n = 30, so the sample is so normal

t =

Conclusion: We reject the null hypothesis at the .05 level because our p-value < α.  Therefore, there is enough significant evidence to prove that the true mean proportion of people that prefer Pepsi is not the same as people that prefer Coke.