Designing a Research Study

1
Designing a Research Study

• The Chocolate Example

2
Research Question

• As a chocoholic, I want to know more about
  chocolate.
• Since that topic is too large, I need to cut it
  back
• Lets use candy bars another topic near and
  dear to my heart
• Even the topic of candy bars is still a little
  broad, so lets cut it down to just one type
• MMs theyre my moms favorite

3
Background Research

• Now that I know I want to study MMs, I need to
  do some background research to figure out what
  type of study I can do.
• Since we dont have time to do a full study,
  well skip the step of reading what type
  research has already been done on MMs.
• Instead, lets take a look at the MMs website
  to see if we can find an interesting topic.
  (http//us.mms.com/us/)

4
Developing a Hypothesis

• Now that weve done our research, what specific
  things could we study about MMs?
• Since Im the lead researcher on this project,
  Im going to make an executive decision to have
  us study the colors in a package of MMs.

5
Developing a Hypothesis, cont.

• Mars says that they want the following color
  breakdown
• Brown 13
• Yellow 14
• Red 13
• Blue 24
• Orange 20
• Green 16
• Remember that we try to disprove things in
  science, so Im going to make the following
  prediction
• H1 The average percentage of a given color of
  MMs in a package will equal Mars' predicted
  percentage for that color.

6
Choosing the Sample

• First, we have to determine the population.
• Next, we have to come up with a good sampling
  design
• Which size of bags to we want to check?
• How many bags do we want to check?
• How will we select the specific bags to check?
• Keeping in mind that there is a cost with each
  bag of MMs, I had to limit us to 1 bag for
  each member of the class, and they had to be the
  candy bar size bags (for cost and our research
  question). Oh, and given the time limitations, I
  think Amber bought all of them at our local
  Wal-Mart.
• This, Im afraid, is the reality of academic
  research.
• We come up with an awesome design, then we have
  to scale it back to what we can actually do.

7
Designing the Method

• Now that weve got our sample, how are we going
  to actually test our hypothesis?
• Remember, we want to know if the percentage of
  each color is within 1 of Mars' predicted
  percentage.
• So, heres our method
• Well each count the number of each color of
  MMs in our bags
• Well calculate a percentage for each color in
  our individual bags
• Well then use a t-test to compare our average
  percentage to Mars predicted percentage

8
Conclusions

• So, what did we find?
• Does that support or disprove our hypothesis?
• Im not sure that it does. You see, Mars doesnt
  claim to get it right all of the time. They say
  that they try to get it within 1 of their
  reported goals. Weve got to check that
  possibility as well.

9
Sensitivity Check

• To make sure that our test actually worked, were
  going to run one more test.
• This time, were going to calculate what are
  called confidence intervals for each color
  Confidence intervals, in this case, are the
  acceptable range for Mars production.
• To get the acceptable confidence interval we use
  this equation
• Reported number /- Std Error T-value
• The T-value comes from the t-table. Usually, we
  use 1.96 (thats what matches up with our 5
  cutoff).
• Hopefully the results will support the conclusion
  we just made.

10
Limitations

• Were the results from our two tests consistent?
• Were there any weaknesses in our sample?
• Were there any weaknesses with our method?
• Is there any weakness with our original
  hypothesis?
• Is there anything else that you can think of?

11
Write it UP!

• At this point, we would probably finish writing
  up our paper.
• Unfortunately, I think that the limitations are
  so severe that we would never get published.
• Instead, youre going to write up your own
  version of a research paper for your final
  project.
• You should follow the same basic method that we
  just went through
• You dont need to gather data or run tests, but
  you do need to come up with at least one test
  that you could run to test your hypothesis.
• As with everything else, please feel free to come
  by and get help if you need it.