Selecting Research Participants
Chapter 5: Selecting Research Participants
The selection of research participants is a VERY important part of planning a research project.
I am going to use YOU (or maybe more properly, YOU ALL -- all students entolled in PSY 3213C OW58) throughout this module as an extended example in order to illustrate many aspects of selecting research participants.
(image of classroom full of students)
Dont worry -- you will never be named by name!
Populations and Samples
A population includes every member of the group of people (or animals) you want to study.
It is important to be clear about who your population of interest is so that you can make sure you have an adequate sample of participants in your study.
Suppose you want to study depression. How will you define your population of interest? Are you interested in all people who have ever been depressed, or only people who are presently depressed?
Suppose you decide you are going to study people who are presently depressed. Are you interested in All people with depression, or only children, or adults, or elderly adults?
Suppose you are interested in elderly people who are presently depressed, and you limit to residents of the United States. Now you have to specify how you will operationalize ‘depressed’ AND ‘elderly.’
Your population will be very different if you operationalize ‘elderly’ as adults over age 60 versus adults over age 70.
Similarly, your population will look different if you operationalize ‘depressed’ as ‘being diagnosed as depressed by a licensed clinical psychologist who has administered the Structured Clinical Interview for DSM-IV” versus operationalizing depressed as “obtaining a score greater than or equal to 14 on the Beck Depression Inventory”.
The ‘target population’ is all elderly depressed persons in the US. The ‘accessible population’ is those members of the population you have access to and from which your sample is selected (see pp. 138-139).
It will be impossible for you to study all elderly persons in the US that are presently depressed. Instead you will study a ‘representative sample’ of persons who are members of your accessible population.
Samples
A sample is simply the group of individuals who actually participate in your study.
Figure 5.2 on p. 140 provides a good visual image of the relationships among the target population, the accessible population, and the sample.
Statistical techniques are based on the assumption that your sample is representative – that means that the sample has the same characteristics as the population of interest (see pp. 140 – 141).
A biased sample is one that does not resemble the target population. Results obtained from a biased sample may not generalize to the population of interest.
Generalization will be discussed further in chapter 6. In short, we say that results are generalizable when we are confident that the same results we obtain with our research sample would be expected if we had studied every member of the target population.
Selection bias or sampling bias, occurs when the sample is selected in a way that increases the chances that it will not resemble the population of interest.
Dewey defeats Truman!
(image of Harry Truman holding up Newspaper with erroneous headline)
The Chicago Tribune printed this incorrect headline after the 1948 presidential election. A sample of voters they surveyed did not resemble the population of voters who elected Harry Truman. Their sample of voters was biased. They conducted a telephone survey and asked a sample of individuals who they intended to vote for. This was a biased sample because in 1948 not everyone had a telephone and those with a telephone tended to be wealthier and to have more stable homes. An individual in this sample was more likely to vote for Dewey than was the average person in the entire population of voters that elected Truman.
The lesson here is that sampling bias can cause misleading results.
Sample Size
A good rule of thumb is ‘the bigger the better’
The larger your sample size the more likely it is to be representative of the population of interest.
Factors such as the time it takes a participant to complete the experiment, the time and effort it takes to recruit participants, the availability of participants, the cost for each participant to participate, the resources available to you, and the type of statistical procedures being used to analyze your data will all be factors that influence the sample size selected.
Sampling procedures
Sampling procedures are also called sampling techniques or sampling methods.
Your textbook describes the difference between random and non-random sampling on pp. 143 - 144. Make sure you understand this distinction, and that you can name which procedures described below are examples of probability sampling and which are non-probability sampling procedures.
Turning to the example of this class, that is the 125 students enrolled in psychology 3213C section 0W58…
Question: is this class a population or a sample?
I hope you answered, “it depends” or ”that is a trick question.”
When you complete course evaluations, or if I determine the average grade on a quiz, then you are a population of interest.
If someone were comparing grades in online versus to face-to-face psychology classes at UCF, or surveying US psychology majors’ understanding of research methods in psychology, or college students’ attitudes towards gun control or marijuana legalization, then you would be considered a sample.
Simple random sampling (p. 144)
In simple random sampling each member of a population has an equal chance of being selected as a participant.
Your discussion groups were created with this type of procedure. A computer program randomly selects students to populate each group. If I wanted a simple random sample of 10 students from the population of students enrolled in psychology 3213C 0W58, then I might use a similar process and ask the Canvas program to create a group or sample of 10 students.
It is simple random sampling when the population is well defined, all members of the population are known, and a sample is selected through a random process.
Question: would the group selection process be an example of sampling with or without replacement?
The answer to this question can be found on p. 145 in your textbook – and you should look it up!
Systematic sampling (p. 146)
Systematic sampling would be used if I created a sample of 12 students in the class by listing all or your names and selecting every 10th student. This may look random, and it isn’t truly as random as systematic sampling, because once the first participant is selected the rest do not have an equal chance of being selected. For instance, if the third student on the list is chosen as sample member #1, then the 13th student on the list has a greater chance of being selected than the 4th and 10th students do.
Stratified Random sampling (p. 147)
Stratified random sampling is used to make sure that the sample includes members of selected sub-groups of the population.
For example, if I wanted to make sure that a sample of students in the population of this class includes women and men, or juniors and seniors, or is ethnically diverse, I would first ‘stratify’ the class into the identified groups and then randomly select participants from each group.
So, if I wanted an equal number of men and women in my sample of 12 students in the class, I would identify all of the women and randomly select 6 of them, and then identify all of the men and randomly select six of them and combine these subgroups for a group of 12 students.
Notice that this is not completely random, because men and women do not have an equal chance of being selected. There are many more women than men in the class, so any man in the class would have a greater probability of being selected than would any woman in the class.
This method is especially useful when you want to compare groups and/or make sure that members of each sub-group are represented in the sample.
The main disadvantage is that the sample may not be very representative of the population. A sample of six men and six women would not be representative of a population of 109 women and 16 men. If I wanted to know the average height, weight, or time spent playing video games for the population, my sample of six men and six women would over-estimate the average for the entire class since men – on average - are taller than, weigh more than, and spend more time playing video games than the average woman.
Proportionate Stratified Random Sampling (p. 149)
This is a variation of stratified sampling that attempts to minimize some of the disadvantages of stratified sampling. A sample of 12 students selected from the population of students in this class using Proportionate Stratified Random Sampling might include 10 women and 2 men, or 9 seniors and 3 juniors, or 6 non-Hispanic Caucasians, 3 Latino/Latina students, 2 African American students and one Asian student. As in Stratified Random sampling, the population would be divided into subgroups, and members from each subgroup would be chosen IN THE SAME PROPORTION as their proportion in the population of interest.
Some drawbacks of this method are that it is a lot of work, and it is some groups of individuals may be left out if they make up a very small proportion of the sample. For instance, in the category of ethnicity described above Pacific Islanders and Native Americans are left out….
Cluster Sampling (p. 149)
Cluster sampling is used when a group, or ‘cluster’, of individuals is selected all at once rather than selecting individuals one at a time.
If I selected, say, discussion groups 3 and 8, as my sample of students in the population of interest – student sin this class – then I would be using cluster sampling, because these clusters of students have already been defined.
The entire class might be regarded as a cluster if the population of interest were “students enrolled in psychology classes in Spring of 2013”, or “UCF students.”
What are the advantages and disadvantages of cluster sampling? (see p. 149 for the answers.)
Convenience Sampling (p. 151)
This is the most common type of sampling even while it has the most problems.
It is common because - as the name implies – it is convenient.
I would be using convenience sampling if I were studying, say, attitudes of Americans towards marijuana legalization, and I posted a note to all of you in the class that said “you can earn extra credit if you complete my survey about your attitudes towards marijuana legalization.”
It would be a convenient way to get data, and the sample would differ from a sample selected RANDOMLY from the population of ‘all Americans.’
A convenience sample selected from this class would be younger and more educated and more likely to be female than ‘the average American.’ Individuals in this convenience samples are also individuals who are more likely to respond to requests to be a research participant than is the average person, which may or may not affect the outcome of the research.
Read about ways to minimize the problems associated with convenience sampling described on pp. 151 – 152.
Quota Sampling (p. 152)
Quota sampling is a form of convenience sampling where the composition of a convenience sample is controlled by establishing quotas.
For example, I could use quotas in the example above to make my sample of students in the class slightly more representative of the population of ‘All Americans’ if I made sure that at least four (or five or six) individuals in my sample were male, or over the age of 25, or were born outside of Florida.
Notice that this is similar to stratified sampling, except it is a form of NON-PROBABILITY SAMPLING, while stratified sampling is a form of probability sampling. The example above is not probability sampling because the individuals that make up the sample do not have the same probability of being selected as other members of the population of interest.
Wrap-up
Note that, in practice, the sampling methods described above are often combined.
See pp. 153-154 for a nice summary of sampling methods.