Page 93 - Design for Six Sigma for Service (Six SIGMA Operational Methods)

74 Chapter Four

owner, the individual consumer could be the unit of analysis; for a mortgage
lending operation, each household could be the unit of analysis; for a
medical equipment supplier, each hospital or clinic could be the unit of
analysis. The population is defined to be the collection of units of analysis
that findings of the survey will apply to. For example, the population of fast-
food chain customers is the collection of all potential individual customers
that the chain can reach; the population of the customers for a mortgage
lending operation is the collection of all the households that the lending
operation could do business with; the population of the customers for a
medical equipment supplier is the collection of all potential hospitals and
clinics that could do business with this supplier.

However, in any population, usually not all the units of analysis can be
identified and reached. For example, if a population is to be the people living
within a metropolitan area, then the unit of analysis will be each single resident.
From a practical point of view, it is unlikely that all the residents of this metro-
politan area can be identified and reached. People are born and people die;
people move in and out. There are people who do not have telephones or stable
living places. Usually only a portion of the population is identifiable and
reachable; this portion of the population is often called a working population.
From the working population, it is possible to develop a list of units of analysis
that can be readily reached in our customer survey. This list is called the
sampling frame. For example, if the population is all the residents in a metro-
politan area, then the working population could be all residents that can be
reached by phone, and the sampling frame could be the residents listed in the
local telephone directory. Some other possible sources for the sampling frame
include voter lists; utility (gas, electric, water, and so on) customer lists; motor
vehicle registrants; magazine and newspaper subscriber lists.

With most sampling frames you will have to deal with some of the following
problems:
1. Missing elements: Legitimate members of the population not
included in the sampling frame. For example, in some polls of U.S.
elections in 2004, only traditional phone users were polled; people
with cell phones only were not selected in the poll list. Therefore, a
sizeable portion of young professionals was left out.
2. Foreign elements: Some people’s names are listed in the sample
frame, but they are actually no longer in the population. For example,
people could have moved out a while ago but their names are still in
the phone directory.
3. Duplicated elements: Population members listed more than once in
the sample frame.