Sampling the Focus Group - Case Study Example

SAMPLING THE FOCUS GROUP Focus groups are comprised of individuals assembled to discuss a particular and differ Nominal groups - researchers do not meet individually with members of an organization Delphi groups - focus groups are not generally made up of trained experts Brainstorming sessions--focus groups do not set out to generate new ideas. Rather, focus groups are designed to gather information from the lay people. The goal in organizing focus groups is to investigate concerns, experiences, or attitudes/beliefs related to a clearly defined topic. In this survey for HCC club we are concerned about the tastes and preferences of the members and other people regarding the services provided at the HCC golf club. Scheaffer, Mendenhall, and Ott (1986) suggest the following nine-step procedure in planning a survey. 1. State the Objectives: In this survey we need to know about the customers taste and opinion regarding the services which should be provide in the club in order to improve the club membership. 2. Identify the Target population: All the customers of the club and the adults of Epson area. 3. Recruiting participants: Recruitment is the process of gathering the group together in the same place at the same time. There are several ways to go about this. Membership lists are a great way to start. Another way is to find a contact that knows the target group. Getting referrals from others, or through word of mouth, is a good means of gathering a sample. If one person is interested, she or he may be able to provide names of other potential participants. This type of recruiting is known as the snowball technique (Lindlof, 1995 as cited by Barnett). Other considerations include demographics. A further question is whether to target a heterogeneous (everybody is different) or homogeneous (everybody is as similar as possible) sample. Most researchers prefer a homogeneous group with the common threads being the issues for discussion (Vaughn, Schumm, & Sinagub, 1996 as cited by Barnett). In this approach it is believed that having too many different voices could detract from the overall purpose. Those advocating heterogeneous groups argue that focus groups should capture a range of opinions, and that participants should feel able to present their perspective free from the fear of appearing different. In a heterogeneous group everybody is, by definition, different. It is important to consider whether focus group members will know each other or whether they will be complete strangers. The degree of familiarity unquestionably impacts group discussions. For the purpose of this research we will consider membership lists and will also contact the Government bodies to get information lists about the people of the area to get idea of people preferences. Since the membership in HCC golf club is very low already, membership lists can support the research purpose but we cannot rely on it. The heterogeneous samples should be selected in order to gain the knowledge about the different preferences of the customers so that the areas should be effectively addressed. It is preferred that group members should not be familiar with one another in order to try to prevent acquaintances from influencing comments. We will select adults aging between 18-49, with different income levels, backgrounds and status in order to get a broader perspective regarding the tastes of the target population. 4. Select a Sampling Frame: After identifying the target population, it is important to obtain a sampling frame. Three problems can occur at this stage. First, it may not be possible to find a perfect sampling frame. For example in the addressed case we get the lists of the residents of Epsom area in order to select the samples for survey. However residents lists may not be complete because they cannot get updated frequently. Even good sampling frames may have omissions. With this knowledge it may be possible to improve the sampling frame. A second problem is called selection error. Increasing the sample size will not solve the problem. The problem is not the sample size, but who is included in the sampling frame. A third problem is called the non-response error. This occurs when a large percentage of the sampling frame does not respond to the survey. Lower-income and upper-income families often do not answer mailed surveys. Thus middle-income families are usually over represented in the final responses. Non response error occurs even in face-to-face interviews. Selection and non-response errors are different. Selection error occurs when the sampling frame does not reflect the target population. Non response error occurs when the respondents do not reflect the sampling frame. In contrast to the selection error, increasing the sample size by re-sampling those who did not initially respond will reduce the non-response error. Furthermore, we should determine whether respondents and non-respondents differ systematically in terms of age, race, or socioeconomic status. To do this, we compare the compositions of respondents to census figures for the sampling frame. If they differ, we cannot draw inferences from the data. Select a Sample Design: The next step is to select the respondents from the sampling frame and determine the sample size. Selecting participants: Most focus group research relies on purposive sampling (Miles & Huberman, 1984 as cited by Barnett), with researchers selecting participants based on the project and on the potential contributions of participants. Alternatively, participants can be randomly selected from a larger group that should be able to give insight into the topic. Selection, response, and non-response errors are mistakes made in conducting a survey. They are correctable. The margin of error is not a mistake. It is the possible difference we allow between the sample result and the result that would be obtained if we sampled the entire population. When drawing inferences about a large population, we must live with a margin of error. At this point, the most effective way to reduce the margin of errors is to increase the sample size. Choosing the appropriate sampling design can also reduce the margin of error. Participant selection requires knowledge of the influence of sampling up on the goals of the group. The strategy of sampling assures the generalisability of the results to the population sample. The following are type of sampling techniques; however, the type of sampling largely depends upon the desired outcomes: There are many sampling techniques such as Convenience sampling, Simple random sampling, stratified random sampling, systematic sampling, and cluster sampling. Each sampling design differs in the method of selecting the sample, the ease of selecting of sample, and the accuracy of the inferences about the target population. Convenience sampling is the selection of participants based upon availability and appropriateness. Occasionally focus groups use convenience sampling (picking people the easiest and fastest way possible) but this strategy is not recommended. Purposive sampling can be broken into specific strategies. Patton (1990) as cited by Barnett, listed five of these: 1. Extreme or deviant case sampling is used to identify a subgroup within a culture. For example, drug users might be recruited for a focus group on a needle exchange. 2. Typical cases provide a cross section of a larger group. 3. Maximum variation case sampling identifies individuals who are able to adapt to different kinds of contexts and conditions. 4. Critical case sampling looks for individuals representing the most "critical" or relevant cases for transfer of findings to other related cases. 5. Politically important or sensitive cases are used to investigate important issues through the use of individuals who have particular viewpoints. It is important to emphasize that regardless of sampling method, focus groups do not provide generalisable results--that is, the findings cannot be applied to all people similar to the participants. The most useful measure of validity may well be transferability, which asks whether the results are presented in a way that allows other educators to judge whether the findings apply in their context. Simple Random Sampling Design: Simple random sampling is a procedure by which, all possible candidates have a fair chance of being selected for the group. Simple random sampling occurs when every possible sample of size n elements has the same chance of being selected from a target population on N elements. The sample is a simple random sample. The goal in obtaining data through survey sampling is to use a sample to make precise inferences about the target population. Moreover, we want to be highly confident of our inferences. We begin with the basic probability sampling design- simple random sampling. A Note of Caution: Probability sampling, such as simple random sampling, is more effective than non-probability sampling designs. For example, in convenience sampling we select the first n customers who take the membership. Although such a sampling scheme would be easy to carry out, we cannot legitimately calculate the probable margin of error. There is no guarantee that the first n customers represent the target population of all customers. A simple random sampling design can also produce very poor estimates. When the sample does not mirror the target population, the estimate is not accurate. Non-representative samples can occur in simple random sampling because each sample has the same chance of selection. Even rare samples, such as those composed of all enlisted personnel or all officers can (but are not likely to) occur. In summary, probability sampling is superior to non-probability sampling. For some situations, simple random sampling is not the best type of probability sampling. We need additional survey sampling designs. Stratified Random Sampling Design: Stratified sampling involves selecting participants so that certain subgroups in the population are adequately represented. The goal in obtaining data through survey sampling is to use a sample to make precise inferences about the target population. Properly done, stratified random sampling can produce more accurate inferences than simple random sampling. Consider stratified random sampling When there are subgroups within a target population that are likely to have similarities –attitudes or tastes– within a stratum but differences among strata. In the case of surveying customers needs to improve membership in the GCC focus group samples can be divided into different strata regarding their age, income and class. Selecting Stratified Random Sample: We begin the process of selecting a stratified random sample by determining the total sample size. Larger samples have smaller margins of error but are more costly. Thus, balancing precision and cost calls for managerial judgement. After determining the total sample size, we must decide on the sample size per stratum. There is much way to allocate the total sample among the strata or groups. Each allocation method may result in a different margin of error. Three factors affect the allocation: 1. the total number of elements in each stratum; 2. the variation in the measurements within each stratum; and 3. The cost of obtaining observations from each stratum. Use proportional allocation sampling When sampling costs are the same for all strata and when all strata exhibit the same variability. A Note of Caution: Properly used, stratified random sampling provides more precise estimates than simple random sampling. However, stratifying may produce worse results than simple random sampling if we stratify the target population incorrectly. Stratifying works best when the variability of the variable of interest is low within each stratum and high between different strata. For the purpose of this survey we will use stratified random sampling in order to gain knowledge about the different factors effecting the membership of the club. Other Designs: Systematic and Cluster Designs. Two other commonly used survey sampling designs are systematic and cluster sampling. Systematic sampling occurs when we select one element at random from the first k elements in a sampling frame, and then select every kth element thereafter. This is a 1-in-k systematic sample. Systematic sampling simplifies the selection process. This design is especially useful when we do not know the target population’s size, and so simple random sampling is not possible. Cluster sampling occurs when we randomly select a set of m clusters from a target population and then examine or interview every element within each selected cluster. The merit of cluster sampling is that it is cost-effective. By sampling all customers within a cluster, we reduce travel costs. Consider cluster sampling when (1) a sampling frame listing population elements is not available or is costly to obtain, or (2) travel costs are very high for other survey designs. Clustering and stratifying are two different methods of grouping a target population. The elements within an ideal stratum should have similar measurements on the variable or interest, but there should be large differences on the variable of interest among strata. As in the case of HCC we divided customers in different stratum according to their age group. The use of Stratified Random Sampling Design helped to find the difference in the requirements of customers according to their age group. In contrast, the element within an ideal, cluster should have widely different measurement on the variable of interest, but there should be small differences in the variable of interest among clusters. References 1. Barnett, J. M., (2005). Focus Groups Tips for Beginners Texas Center for Adult Literacy & Learning Available http://www-tcall.tamu.edu/orp/orp1.htm 2. Brightman, H., Schneider, H., (1994). Statistics for Business Problem Solving, (2nd ed.). SouthWestern Publishing Co., Ohio. 3. FIE-SDFS, Research & Evaluation, Designing A Successful, Focus Group, available http://www.unf.edu/dept/fie/sdfs/research/focus_groups.ppt 4. Lindlof, T. R. (1995). Qualitative communication research methods. Thousand Oaks: Sage. 5. Miles, M. B., & Huberman, A. M. (1984). Qualitative data analysis: A sourcebook of new methods. Beverly Hills, CA: Sage. 6. Patton, M. Q. (1990). Qualitative evaluation methods (2nd ed.). Newbury Park, CA: Sage. 7. Scheaffer, Richard, William Mendenhall, and Lyman Ott. (1986). Elementary Survey Sampling. Boston: Duxbury Press. 8. Vaughn, S., Schumm, J. S., & Sinagub, J. (1996). Focus group interviews in education and psychology. Thousand Oaks: Sage. Read More