Research Methods

Research methods are the tools qualitative researchers use to investigate their research topic and construct their argument and the decisions they make as to how to use those tools and with whom. As noted earlier, qualitative research methods share a common core of characteristics. They are generally used in face-to-face situations in which the researcher is relating to the respondent or the setting or both. The researcher is the primary (though not the only) tool for data collection, meaning that information is always filtered through the exchange between the individual, the research setting, and the respondents. This may introduce biases into the interview process. To try to reduce researcher bias and to enhance the voices and interpretations of respondents, while supporting researcher engagement, qualitative researchers attempt to minimize personal characteristics that could interfere with communication. Doing so requires researchers to reflect constantly on how they may be influencing the research setting and the research conversations by virtue of their identity, language capacity, knowledge of local culture, customs and etiquette, and perceived power or access to resources desired by the respondents. It also requires researchers' careful contemplation of their own possible biases or strongly held attitudes about local practices or people in the research setting that could wrongly influence interpretations or understandings of the field situation.

Two strategies researchers use to reflect on these issues are keeping a diary or personal log and creating an initial or formative conceptual model. The diary allows the researcher to record experiences in the field that are cause for reflection and consideration and that may require a change in approach or communication style. The conceptual model uses logical arguments drawn from the literature and personal experience in the study site to identify and link the main domains that are believed to be the most important for the study at initiation. Producing a conceptual model requires clear explication of both the reasons for choosing domains and the links among them. For example, the initial conceptual model for a study of HIV exposure among older adults consists of an outcome—exposure to HIV—and the factors believed to contribute to it, including the presence of injection drug users in their immediate networks, the presence of drug dealers in the neighborhood, and the involvement with drug-using commercial sex workers (see Figure 4.2) (Radda, Schensul, Disch, Levy, & Reyes, 2003; Schensul, Levy, & Disch, 2003). This model can then be expanded to include other factors as qualitative data on HIV exposure are collected and analyzed.

As another example, a researcher who proposes that inadequate school facilities, poor educational instruction, and a program of rental evictions result in poor school performance is biased in favor of structural or systemic explanations. The resulting model leaves out such factors as low parental education levels and peer norms favoring frequent cell phone texting. Thus a conceptual model will quickly reveal what domains researchers think are connected and should be explored in a study, where they have more or less information already, and what their biases and gaps in knowledge might be. In the early stages of a study, these can be corrected through validity checks carried out by asking local experts and other researchers who know the study setting whether the researcher's initial model and explanation for it make sense.

FIGURE 4.2 Initial Conceptual Model of HIV Exposure Factors Among Older Adults

Although qualitative researchers collect many different types of data to answer their research questions, all types of qualitative research take an emic perspective, focusing on learning and understanding the perspectives of local residents and experts. The emic perspective is based on the belief that people's viewpoints, when set in the context of their lives, are understandable, whether or not the researcher agrees with them. The meanings that people attribute to their actions and behaviors, whether communicated directly or indirectly, are considered central to qualitative inquiry. Once the data are collected, the researcher can determine how and in what ways to represent the voices of the study participants. Always, however, qualitative researchers keep in mind that “sense-making through the eyes and lived experience of the people is at the heart of good qualitative research” (Schensul, 2008, p. 522).

It is useful to organize the collection of qualitative data into a four-cell matrix (see Table 4.2). The matrix juxtaposes two different primary ways of collecting face-to-face data, observation (what is seen and recorded by the researcher) and interviewing (what is told to and recorded by the researcher), against two different primary ways of organizing the data, the cultural level (including information about the community, organization, or collective cognition) and the individual level (including data obtained from individuals about individual beliefs and behaviors).

Table 4.2 Main Classes of Data Collection at Individual and Community Levels

Researcher Position in Data Collection

Researchers may choose to be more or less engaged in the study site, or with the study participants, depending on the paradigms that drive their work and the requirements of the study. In addition, observations and interviews may be more or less exploratory or structured beforehand. The more the researcher prestructures or predefines either observations or interviews, the more focused and limited the nature of the collected data. Observations may be more or less obtrusive (intruding on the regular lives and behaviors of those who are the subject of study) depending on the degree to which the researcher structures the activities to be observed. For example, observation in a kindergarten classroom to document and describe predetermined types of interactions among teachers, aides, and students requires that the researcher become familiar with the study setting so as not to attract attention or interfere with the regular daily schedule. In this type of observational setting, the researcher can observe everyday behavior, document interactions, and define and code or classify them in the research site or later on. Conversely, experiments, in which respondents are asked to perform an activity and researchers code observations of the activity for the presence or absence of types of behavior—or for emotional responses that are derived from the study's theoretical framework—call for a high degree of structuring of both the behavioral setting and activities. A typical example might involve asking parents to use a standard set of developmentally appropriate toys in the same setting to demonstrate how they play with their toddlers.

In the same way, the more specific and structured the interview schedule, the less opportunity respondents have to express themselves as they would in a regular conversation. Semistructured, open-ended in-depth interviews in which the interviewer is required to follow a sequence of questions limit the exploration allowed, as compared to more individualized and unstructured—or minimally structured—open-ended in-depth interviews. Further, the time limits imposed by requirements to ask all the questions in an interview schedule may reduce the possibilities for the researcher's engagement with the participant. At the same time, very open-ended interviews—which are conducted more like regular conversations—require the researcher to have considerable skill in focusing the questions so as to collect useful information relevant to the study and keep the respondent engaged.

Data Collection at the Cultural Level

At the cultural or collective level, qualitative researchers try to obtain information on community- or systems-level phenomena. These might include activities in which many or most people in the community participate, such as annual meetings, festivals, and religious celebrations that can illustrate community dynamics and tensions as well as cultural practices. Events that mark turning points in the history of a community or a group of residents also offer important information about the present. For example, stories obtained from residents of New York about the attack on the World Trade Center (WTC) buildings can provide a context for people's feelings about and behaviors in response to the placement of a Muslim cultural center near the WTC site. Further, in studies of drug use in Hartford, interviews about the destruction of public housing and the disbursement of public housing residents throughout the area provide a lot of information about the decentralization of social networks and the spread of specific drugs from the suburbs to the city and vice versa. In order to gather these types of data on community-level activities, beliefs, and interactions, the primary data collection tools are observations and interviewing. These are the building blocks of qualitative data collection.

Qualitative researchers may use other data collection methods to obtain information at the cultural level. These could include any or all of the following:

• Cultural consensus modeling, which provides information about the components or elements in cultural domains (such as leisure time activities, types of risk, illnesses, types of clothing, foods) and the different ways the way people organize and classify them; how people explain these mental or cognitive groupings; and the degree to which there is consensus or agreement about the ways the items in a domain are grouped
• Network research, which involves documenting through observation and measuring with surveys the ways organization members and organizations or specific locations, such as bars, libraries, or senior centers, connect to each other, in what ways, and for what reasons
• Archival research, which involves using secondary data (primary data collected by others but available to the public for use) or library source data to help understand the history of a study site
• Community mapping in various forms, including drawing maps of the community or asking residents to draw such maps, and using existing to-scale maps or Google maps to locate activities and organizations spatially in relation to where people live and conduct their daily activities
• Audiovisual documentation, which involves filming or audio-recording activities that take place in the community for later coding and analysis

All of these methods round out the community-level data collection repertoire and provide the basis for a detailed description of a community or a study site.

Data Collection at the Individual Level

At the individual level, the focus is on the discovery of main themes and range of variation in the experiences, beliefs, norms, and practices of individuals. Again the primary means of data collection include interviews and observations, both open-ended and structured. These interviews are gathered from more than one person, because the goal is to identify differences and similarities across respondents in a sample. The following are the main ways of collecting data from individuals:

• In-depth interviews are conducted with unique individuals or a small number of people. There are several types of in-depth interviews. Life history narratives involve few interviews, which are usually very lengthy (up to fifteen or more hours of interview time); narrative interviews focus on specific and often sensitive topics, such as bereavement or HIV, and usually consist of three interviews of about one to two hours each, moving from less sensitive and more descriptive to more sensitive and more focused on personal meaning and feelings. One-time in-depth interviews usually address a specific topic and last about one to two hours (Seidman, 2006).
• Semistructured interviews are used to collect similar information from a larger sample of individuals, numbering at least twelve to fifteen and usually not more than ninety.
• Qualitatively based surveys are based primarily on prior qualitative research in the study population. These surveys are generated from the domains, subdomains, and individual items that emerge from in-depth and semistructured interviews. Usually they do not include standardized scales and other validated instruments, although there is no hard and fast rule about such inclusions. However, if preselected scales are used, it is always best to pilot them for meaning as well as to analyze the structure of these scales to make sure they are internally consistent. There is a strong possibility that any standardized measure will require adaptation when used with a new study population. The same principle applies to a standardized behavioral coding scheme, which will require adaptation to the study situation and setting.
• Individual-level network data (ego-centered data) describing the personal networks or relationships of individual respondents in a study can be collected, even in the context of in-depth interviews. Person-oriented network research can show what proportion of an individual's network members are involved in risk behavior (which is a more specific behavioral indicator of social influence than perceived influence). These data can also show what proportion of a personal network provides support for or extracts support from an individual (these are measures of positive and negative social support).

  A number of data collection methods or tools can be used at both the cultural (community or organizational) level and individual level for network research. For example, a researcher can collect network data to understand the structure of relationships among individuals in a bounded system, such as a classroom or buildings in which older adults live. In these cases it is possible to ask every member about every other member. In a semibounded system, such as a kinship network or peer network, the cutoff points demarcating network members may be unclear. It is then not possible to collect information from each and every member of the network about his or her relationships with all the others, so the network data are incomplete.

Most in-depth interviews or narratives can be coded, compiled, and even quantified to illustrate both the range of variation and cultural themes and patterns across individuals in a setting.

It is important to note that each of these approaches to data collection requires a design for site selection, observation, and recruitment. Discussions of recruitment strategies can be found in a variety of sources on qualitative methods (for example, see Bernard, 2000; Schensul, Schensul, & LeCompte, 1999).