The Coors Challenge

1
The Coors Challenge

• Alex Dinn
• Larry Hedden
• Scott Kennedy

2
Introduction

• According to anecdotal accounts, many believe
  Keystone Light and Coors Light share a similar,
  if not indistinguishable, taste.
• However, we believe that most will in fact prefer
  the taste of Coors, considering it costs
  approximately 40 more.
• Our belief going into the experiment was that
  approximately 72 would choose Coors over
  Keystone, reflecting the price difference.

3
Procedure

• We asked participants to sample two cups of beer
  and asked the following two questions
• Question 1- Can you tell the difference between
  the two samples?
• Question 2- Which of the two cups do you prefer
  the taste of?
• Note- Our first question was designed to isolate
  those that could taste a difference since that is
  the population we are interested in (ie
  discerning tasters)

4
Our Null Hypothesis

• Our null hypothesis states that P0.5, where
  Pnumber of tasters who preferred Coors/number of
  discerning tasters.
• Thus, the null states that test takers are
  equally likely to choose each type of beer.
• Our alternate hypothesis states that Pgt0.5.
• To reject the null hypothesis, we needed a
  critical value of p0.62 for 95 confidence
  level.
• Thus, we are leaving a 5 chance of a type I
  error, that is we reject the null when its true.

5
Results

• 30 out of 51 tasters chose Coors. Thus, the test
  statistic we obtained was p0.59.
• Therefore, we fail to reject the null hypothesis
  since our test statistic is below the cutoff of
  0.62.
• Assuming the null is true, we would expect of
  value of a test statistic this extreme around 13
  percent of the time due to chance.
• Thus, we suspect our null might not be true but
  we cannot reject the null with statistically
  significant results.

6

7
Power

• According to our belief that 72 of people would
  select Coors, we calculate a power of
  approximately 95 given our sample of 51 people.
• We have a 5 chance of a type II error, which
  means that we fail to reject the null when in
  fact it is false.
• In this case, we decide we would determine test
  takers are equally likely to choose each beer
  when in fact they are more likely to choose
  Coors.
• Given that our test statistic is just below the
  cutoff, power is very important.
• Our results lead us to believe that the true of
  value of P might lie somewhere in between 0.5 and
  0.72.
• Interestingly, if our null hypothesis were
  p0.72, then we would expect a value as extreme
  as 0.59 only 3 of the time.
• So were fairly certain the true value of P is
  less than 0.72 but perhaps we would need a larger
  power to detect such a subtle difference.

8
Issues that affect our results

• Temperature of the beer
• Amount of beer given to participants
• A choice of no preference- possibility of asking
  respondents to repeat the process
• Should we have told the test takers that they
  were sampling Coors Light and Keystone?
• Possibility of a control group- two of the same
  beer given to a test taker- might be difficult to
  interpret data?
• Randomness of population questionable- people
  over 21
• Importance of Greek affiliation and gender- 57
  percent of those in fraternities chose Coors and
  52 of women in our sample chose Coors, leading
  us to believe that the effect of these
  demographic features might be minor.

9
Conclusion

• From our results, we fail to reject the null
  hypothesis at the 95 confidence level.
• We conclude that we do not have enough evidence
  to show that there is a statistically significant
  difference between the tastes of Coors and
  Keystone.
• However, again since our test statistic is fairly
  close to the cutoff, we have reason to believe
  that with a larger and more varied sample size,
  we might reject the null (ie drink more beer).