Question

An airline is planning on making changes to its Frequent Flyer program. Before implementing them, they want to know what their customers think about the changes. They plan to sample a portion of their customers who are members of their Frequent Flyer program. They have two lists:

• One sorted by what level of frequent flyer the customer is: Gold (15 or more flights a year; about 15% of customers), Silver (5 to 14 flights a year; about 25% of customers); Bronze (less than 5 flights a year; about 60% of customers)

• One sorted by city the customer normally flys out of: There are 15 Canadian “hubs” or cities on this list

1. (a) Which of the two lists would you use if you plan to use Stratified Sampling and why? Then give 1 or 2 lines explaining how you would use the list to get your sample of customers. (1 + 1 = 2 marks)

2. (b) Which of the two lists would you use if you plan to use Cluster Sampling and why? Give 1 line explaining how you would use the list to get your sample if you use “single stage” clustering. Then give 1 line explaining how you would use the list if you use “multi - stage” clustering. (1 + 1 + 1 = 3 marks)

Answer 1

a] We would use the list of level of frequent flyer customers, if we plan to use Stratified sampling.

Here the strata will be Bronze, Silver and Gold. Then a random sample is drawn from these strata.

Stratified sample is obtained by dividing a population into non-overlapping, homogeneous groups called strata, and then selecting a random sample from these strata. The groups Gold, Silver and Bronze used in the list satisfy the condition of non-overlapping and homogeneous. Hence this sampling procedure is apt for the given list.

b] We would use the list sorted by city the customer flys out of, if we plan to use Cluster sampling.

If we use single stage clustering, the random sample of clusters will be the 15 Canadian cities.

If we use multi-stage clustering, the 1st stage will consists 15 cluster of Candian cities. The next stage may include clusters of Urban and Rural areas of each city.

Cluster sampling is used when natural but heterogenous groups are present in the data. The common variables used in clustering population are geographical areas, buildings, etc. Here the list consists of natural but heterogeneous groups of cities and hence this sampling procedure is apt.