Scenario

The data you have been given have been generated to match the scenario described in Supercrunchers Page 58 which explores the experiment ran by Kelly Cook in order to study how best to handle inconvenienced passengers. You will consider this as an example of an A|B test. In the book, they do however talk about two treatment groups and one control group.

Here are the key points of the scenario:

• The passengers examined are all frequent flyers who were subject to a ``transportation event" (that is they had a flight severely delayed or cancelled)
• Following this event, passengers were randomly allocated to one of the following three groups:
  1. A letter of apology
  2. A letter of apology and trial membership of Continental's Presidents Club
  3. Nothing
• Following this, we record data on the value of tickets purchased by these passengers in the subsequent 12 month period.

Your tasks

You have to complete the following tasks and write a brief report of no more than 500 words. You are to submit your report and Excel workings to the course DLE by the deadline of noon 14th March 2016.

You have been supplied with a smaller dataset whereby a sub-sample has been taken at random from two of the three groups. You therefore have data on the following:

• Delayed passengers sent a letter (the treatment group)
• Delayed passengers who received no intervention (the control group)

This has been done purely to make the Excel calculations quicker and easier.

1. Summarise the data for ticket sales for each of the two groups. You should use numeric data summaries (measures of location and spread) as well as those visual displays you deem appropriate. Given that we are working with a subset of the data explain what procedures were followed in order to make this valid.
2. Use a ``Randomisation test" in Excel to determine whether there is evidence that sales in the treatment group (those who received a letter of apology) were higher than those in the control group. You should clearly state the null hypothesis, indicate whether you reject or fail to reject the null hypothesis and describe what this means in terms of the effectiveness of the intervention. You should also explain who you think the ``population'' is for this study to whom you may generalise the results (for example would the results to all flyers, all frequent flyers and so on).
3. A colleague suggests you should create a Monte Carlo simulation to explore the likely cost-benefits to the airline if you apply this intervention (send a letter of apology) to all delayed/cancelled frequent flyers. Carefully explain how you would set up such a study, naming the random variables you would use for ticket sales, the parameter values you would select for these random variables, the number of random variables you should simulate and the number of replicates you would run. Explain briefly how you might estimate the cost of the intervention.

Price: $13.11

Solution: The downloadable solution consists of 8 pages, 511 words and 2 charts.
Deliverable: Word Document