Explain the differences among the following sampling techniques, simples random sample, stratified sample, systematic sample, cluster sample, multistaging sample, and convenience sample. and describe a situation each one might be useful.
(1) Simple random sampling -- every element of the population has an equal chance of being selected. One way to achieve this is to assign each individual a number, and the randomly select numbers.
One of the problems with this method is that you need access to a truly random process in order to select the individuals.
You might use this with a relatively small population. For instance, the ATF might randomly select 50 Wal-Mart stores for a gun audit.
(2) Stratified sample -- you divide the population into strata; naturally divided groups. Then you randomly select individuals from each strata.
For a business to gauge the effectiveness of a new policy they might interview random people from upper management, human resources, middle management, accounting, etc...
A school system might interview random superintendents, principals, teachers, office workers, cooks, bus drivers, custodial staff, and students.
(3) Systematic sample -- assign each member of the population a number and then systematically select members to interview. E.g. select every 5th, 1000th, 10000th person from a list.
You do not need to have access to a truly random process to select interviewees. However, the group selected is not random -- each individual in the population does not have the same likelihood of being selected.
This is a cheaper way of generating a sample. For instance, you might call every 100th person in the telephone directory.
(4) Cluster sample -- the population is divided into naturally occurring clusters, then clusters are randomly selected. Each member of the selected cluster is interviewed.
If Wal-Mart wants feedback on a new policy, they might randomly select 20 stores and interview everyone in each of the selected stores.
(5) Multi-stage sample -- a complex form of cluster sampling. First you divide the population into clusters and randomly select some clusters. Then you further divide the selected clusters into smaller clusters. You might repeat this a number of steps until you arrive at a manageable sample size.
A polling group might randomly select House districts. Within each of the selected districts they might randomly select townships. Within the randomly selected townships they might randomly select certain blocks, etc...
(6) Convenience sampling -- This is the cheapest and easiest form of sampling. It is also the least reliable.
A marketer might interview customers entering or exiting a store on theie buying habits or about a particular product. There are many problems with this sort of sampling -- the samples are not random, only a select segment of the population can be interviewed, and their is potential for bias on the part of the interviewer.
For political polls you have to be very careful of your selection technique. Pollsters take into account whether a person is registered to vote, whether or how often they vote, race, income, and many other factors. Pollsters cannot rely on simple techniques like systematic sampling -- for instance you cannot call every 100th person in the directory as you will miss people with unlisted numbers, miss people without landlines, and you will connect with people who do not or cannot vote.
Simple Random-Everyone in the population has the exact same chance of being selected. This could be used for any situation, though it's simplicity means it's more bias.
Stratified Sample-Population is divided into stratas, which are basically subgroups, with each subgroup containing a common characteristic. This is used to ensure the proportions in the sample match that of the population. For instance, if you survey a classroom that is 75% female, then three-fourths of the people sampled should be female as well. This is to ensure homogenity in the population and sample.
Systematic Sample-It's similar to random sampling, except there's a method to picking people. For example, each nth person in a list of names is selected. This might be used for polling for an election, as someone could pick every 10th name in the phone book or something along that line.
Cluster Sample-A population is split into groups, or clusters, and groups are selected at random. Everyone within the selected groups are then surveyed. This is done when the cost or time to survey people within different groups are difficult. For example, if someone wanted to get an idea of the literacy rate in elementary schools in America, rather than surveying some schools in every state, it would be easier just to randomly pick a few states and survey every school in those cities.
Multistaging Sample-This is a more complicated form of cluster sampling. After dividing the population into clusters, those clusters are also divided into another cluster. Using the example from above, the states could be divided into cities to be surveyed. This is more useful for time and cost reasons.
Convenience Sampling-As the name suggests, this consists of getting a sample that is most convenient. Due to this, it has a lot of bias to it. An example would be standing outside a mall and surveying anyone who walks by.