Interviewing representative samples is one way to create generalizable conclusions. The trick to being able to do so, however, is to ensure that your sample is representative. We can do this through two strategies: randomization, which is the strategy used by survey sampling, or deliberate design, which is the strategy used by focus groups.
Random selection of observations – individuals to study – ensures that the selection is not correlated with any factors of interest to us. Why? Because the selection criterion by design is not correlated with anything. Survey firms use complex stratification schemes to obtain fully representative but still random samples. We can use simpler strategies to produce smaller random samples that will suffice for student research projects. The easiest randomization strategy is to sample every nth person – every fifth person, every seventeenth person, whatever is appropriate based on the volume of individuals passing through your survey location, how many observations you need, and how many interviewers you have available to help ensure that you hit all the selected people. (This is one of those times when enlisting a friend’s help in exchange for coffee or ice cream is definitely a good use of $5.) What number you choose doesn’t matter so long as you stick with it and apply it consistently. You will end up with a random sample that can be used to draw inferences about the population from which the sample was drawn. If you were sampling students outside the campus dining hall, you can only generalize to students. If you sampled voters exiting a polling station, you can generalize to all voters (or at least all voters in your community or state). Be aware that you often need permission to do polling on private property as it’s usually considered solicitation; asking for permission even if it’s not officially required is always a nice courtesy. Special rules often apply to polling outside polling stations, so check with your local board of elections to find out the rules in your area.
Focus groups, on the other hand, study a smaller group of people more deeply and with more flexibility in question than a standard survey forma. They attempt to deliberately design a sample that reflects the population of interest. They draw on basic demographic characteristics of the population to craft proportions for the focus group: percent female or minority or conservative or whatever. They then use screener surveys to identify individuals who help them meet the various quotas for each group and invite those individuals to participate. Typically, a series of focus group sessions are held, each with a different panel of subjects, who when pooled together make a complete and descriptive sample. Marketing researchers have developed focus groups into an art form. Two good resources are available at http://www.eiu.edu/ihec/Krueger-FocusGroupInterviews.pdf and https://assessment.trinity.duke.edu/documents/How_to_Conduct_a_Focus_Group.pdf .
PARTICIPANT OBSERVATION
Participant observation as a form of data collection is an ethically challenging procedure. To perform participant observation, one participates in the activities of some group – a meeting, a protest, another function – as if one is a true member of the group. No one in the group knows that you are not there as a ‘true believer’; in other words, you by definition engage in deception when you employ participant observation as a data collection method. Unlike most other situations where researchers employ deception with human subjects, it is often difficult if not impossible to debrief participants (i.e., explain your deception) and/or get informed consent from them. This is why participant observation should NEVER be conducted without thorough review from your school’s human subjects ethics committee. It may be possible with faculty review in utterly anodyne situations such as observing seating patterns in the student cafeteria, but even there, ethical considerations must predominate. For your own safety as well as the safety of those you’re observing, please work with your human subjects committee to get full approval for any participant observation activities you may consider.
As a method of data collection, participant observation (PO) requires you to participate in the group’s activities. Sometimes the goal of PO is simply to establish the range of activities the group engages in, or to observe the procedures or organizational structures used to support those activities. Other times, it requires sustained participation over time and building rapport with members to get them to accept you as a member and begin discussing their beliefs or opinions with you in a way that will not blow your cover. Your goal is often to act as an insider and get inside the heads of the participants, to come to understand their actions as they themselves understand them.
Each PO data collection session should be followed, as immediately as possible, by a session of writing down your field notes from the experience. Begin with the factual basics of what happened in what order, who was involved, who said what, etc. Then add your interpretations, personal notes, and observations/inferences about what you observed: power dynamics, coalitions, bases for justification in debate, whatever is relevant for your research question. A great guide to field notes is at http://www.qualres.org/HomeFiel-3650.html.
Entire books have been written about participant observation. Some other good online resources for beginning participant observation include:
- Basic orientation to participant observation: http://www.write.com/writing-guides/research-writing/research-process/participant-observations-choosing-your-observation-methods/
- Forum: Qualitative Social Research Participant Observation as a Data Collection Method, Barbara B. Kawulich, http://www.qualitative-research.net/index.php/fqs/article/view/466/996 Volume 6, No. 2, Art. 43 – May 2005
- Field guide for PO: http://assessment.aas.duke.edu/documents/ParticipantObservationFieldGuide.pdf