Strategies for Selecting Participants
We will begin with some rules of thumb that guide the selection of participants for focus group interviews. These rules are then woven into several strategies for identifying participants. First, we offer these general rules to guide the selection process.
Set Exact Specifications- The Screens
As precisely as possible, identify the demographic and observable characteristics of the people you want in the group. These characteristics are called the "screens," Here is an example of screens: A participant must be a woman (Screen 1) from Dakota County (Screen 2) with a baby under the age of 1 (Screen 3) and who is a first-time mom (Screen 4). Be cautious when making selections on nonobservable factors such as attitudes, opinions, or values. Use nonobservable factors only if you have nonbiased empirical data with which to make your decisions.
Maintain Control of Selection Process
The researcher should maintain control of the selection process. At times, it is advisable or even necessary to let others make decisions on the selection process or to carry out strategies used for selection. When others are doing the recruiting, they should fully understand the purpose of the study and the strategy they should use for selecting and recruiting participants. Give precise directions. For example, suppose the Postal Service wants to conduct focus groups with experienced mail sorters and clerks in five major cities. Line supervisors will need to be consulted because of the impact on work floor productivity when workers are absent. These supervisors must be aware of and approve the release of the employees. Because of budget limitations, the researcher may not be able to make advance site visits to conduct the employee screening and make final decisions about participation. In these cases, the researcher will need to depend on the local site supervisors to ensure that a sufficient number of participants are correctly screened and able to leave the work floor during the focus group interview. Unless precise instructions are developed for selection and recruitment, there is a risk that those attending will not be typical of the employee category.
Use the Resources of the Sponsoring Organization in Recruiting
A strategy that has been beneficial in public and nonprofit organizations is to use the skills and strengths of the organization to recruit participants, using carefully laid-out protocol developed by the researcher. Suppose a college wanted to conduct focus groups with alumni to discover the ways alumni preferred to be informed about developments at the school. Also suppose that the budget is tight and the college could conduct focus groups with additional categories of alumni if the alumni office could conduct the recruiting. The researcher might want to use the resources of the alumni office-the class lists by year with demographic data to screen the participants, the clerical staff to make the telephone contacts, and the name of the school and the alumni office to establish credibility and legitimacy. Although a research firm could do all of these tasks, the costs would be greater. However, the researcher must be explicit in laying out the steps needed for selection, giving instructions to staff about how to make the telephone request, and preparing the official letter of request.
Beware of Bias
Selection bias can develop in subtle ways and seriously erode the quality of the study. Here are some examples:
Beware of participants picked by memory. Memory is limited and selective. Those names that can be readily recalled may differ in substantial ways from the available study population.
Beware of participants picked because they've expressed concern about the topic. The conscientious supervisor inquires about the purpose of the study, and after it is explained, several names quickly come to mind. These names may include those who have expressed past concern, anger, or frustration with the topic. The supervisor assumes that the study would be improved if it received this input, and also it would show the employees that the supervisor took their concerns seriously enough to suggest they be included in the group discussion.
Beware of participants picked because they are clones of the supervisor doing the selection. It is a human tendency to believe that those who think like we do are dazzlingly bright. Their logic, vocabulary, and values make sense. The supervisor, with the best of intentions, unwittingly selects those with similar views.
Beware of participants who are out of the mainstream. In certain situations, the supervisor may not be interested in releasing the most productive employees to attend a focus group. If someone has to go to the focus group interview and the pressure is on the work crew, it's tempting to send the deadwood. Although it may not be your intent, you may have inadvertently assembled the disinterested employees or those who are tuned out to the organization.
Randomly Select From Your Pool
Randomization helps ensure a nonbiased cross section, essentially giving everyone in the pool an equal chance of selection. However, randomization works only if your pool of prospective participants meets your selection criteria. Randomization is rarely done of the entire population but rather of those passing the selection screens. As a result, even though we randomize to limit bias, we may in effect have the wrong screens.
Balance Cost and Quality
Several factors increase recruitment costs: too many selection screens, selection screens based on nonobservable factors, potential participants who are hesitant to participate, studies that aren't explained well, the sponsor's image, interview locations or times that are inconvenient, or when the benefits of participating aren't clear, just to name a few. In virtually all situations, there are multiple options for recruitment- each with differing costs, efficiency, and quality of result. Quality refers to the ability to locate the right people. At times, compromises are needed, such as finding alternative locations or dropping some selection screens. Creative alternatives can sometimes be found through brainstorming.
Nonusers Can Be Difficult to Locate
A number of organizations have tried to get views of nonusers, but some found that recruiting is quite difficult. Typically, there are no reliable lists of nonusers. Even lists of the population are of limited value if the participants are unaware or uncertain of whether they use the product or service. In some ubiquitous services, such as Cooperative Extension, we regularly find that residents do not consider themselves to be users-but in fact, they have made use of information provided by that organization. As a rule of thumb, nonusers are harder to find, often because reliable screening questions are cumbersome, and in some situations, the participant is just not aware of their use.
Users May Differ in Ways That Can Affect the Study
When organizations, either public or private, seek insight from their users, they often discover that these users differ in frequency or intensity of use. Whether your organization provides social services, information, or breakfast cereal, some will use it more often or more in-depth than others. Does this matter for your study? For example, in a study, church members were categorized in two ways: by how frequently they attended church services as well as how much money they contributed annually to the church. Financial contributions are an imprecise measure of intensity, but in this study, the intent was to determine the level of financial support for a new building project- and financial support was essential if the addition was to be feasible.
No Selection Process Is Perfect
We make the best choices we can with the knowledge we have available at the time of the decision. Selection is limited by our human capacities. We may overlook certain aspects of the problem and inadvertently neglect individuals with unique points of view. A test of the selection process is whether you are able to successfully defend the selection process to colleagues and clients. Trade-offs occur constantly and require weighing a possibility of bias or perception of bias against costs.