CHAPTER 3: METHODOLOGY
3.1 Research method and design appropriateness
3.1.2 Case study methods
A variety of methods are appropriate in a case study research design. Common methods of case study are observation, interview, artifact/document collection, coding, data management, and interpretation (Stake, 1995; Solomon, 2014). In a confirmatory study, instrumentation is useful for comparison with other previous studies and for managing the data across multiple cases. Miles and Huberman (1994) describe three levels of qualitative-quantitative data linkage: quantizing, in which qualitative information is counted; linkage between distinct data types, in which qualitative data is compared to numerical data; and multimethod design, which may include interventions and experiments. This study used the second level, linking data types, to compare qualitative data from observations, interviews, and documents to middle managers’ numerical scores on a questionnaire. Observation provides an opportunity to collect data from informants who are in their natural setting. In this case, the natural setting was the academic library, and the subjects are the middle managers, their bosses, their peers, and their direct reports. Generally, observation may be direct or passive, obtrusive or unobtrusive; the observer may be a participant in the setting who takes part in activities and/or has explicit relationships with the subjects, or the observer may observe from afar and not
take a role in the setting. Participant observation has been broadly defined to include the explicit placement of the observer in the setting, with interaction with the subjects, without the observer having work responsibility in the setting. Methods employed in this study to gather data included observation, interviews, and document analysis, the strengths and weaknesses of which are explained below.
There are several strengths of observation as a method for data collection. It allows the researcher to focus on interactions among subjects, including non-verbal behaviors, that could not otherwise be known. The researcher may also become aware of contextual clues that explain or enhance the understanding of the phenomena being studied. The behaviors that the researcher observes may confirm previously thought ideas or conclusions, and the researcher may observe the values of the organization as they are enacted. The researcher does not need to rely on the subjects to recall behaviors, but can observe them in situ.
Observations have a number of weaknesses as well as strengths. First of all, access to the site may be difficult to achieve; organizations may not want to grant access or may want to control access to individuals and influence outcomes; individuals may not want to be observed, or their behavior may change because they know they are being observed (however, this effect may weaken over time). Granted access, especially in the case of participant observation, subject/observer interaction may influence both the behaviors of the subject and the observer, causing different behavior than might occur if the observer were not present. An observer cannot see everything, even if using a recording technology, and may miss behaviors that are important to the study. Even observations scheduled in advance to capture particular situations may be thwarted by observer effects or otherwise missed opportunities. Many elements are hidden or implicit, and the observer has to take care regarding the misinterpretation of participants’ words and actions.
Interviews allow for in-depth, one-on-one data gathering where the subject describes phenomena from their point-of-view. Interviews may be structured, semi-structured, or unstructured. In this study, interviews provided a mechanism for talking with informants in private settings, allowing the interviewees to confidentially express themselves, generally without fear of exposure.
Like observations, interviews have inherent strengths and weaknesses. Semi-structured interviews, with the same set of questions given to each employee type, allow the researcher to gather basic demographic, behavioral, and cognitive data as well as hear the point-of-view of the participant
in their own words. Given adequate probes, the participant might reveal unexpected information that moves the research forward and helps the researcher understand the setting. Data from similar interviewees (in this study, the same employee type), can be analyzed for commonalities and differences in how they talk about or otherwise characterize the phenomena being studied.
With interviews, weaknesses abound. The questions have to be carefully designed, with appropri- ate probes, in order to get adequate responses from the participant. In the case of semi-structured and unstructured interviews, the skill of the interviewer is extremely important, as well as the establishment of rapport between the interviewer and the respondent. The free-flowing data from these interviews present coding challenges, and some of the unexpected information, noted as a benefit above, may be unwelcome and send the interview into unwanted territory. The researcher is also at the mercy of the individuals in the organization, many of whom may not want to be interviewed or may not be consoled by the promise of confidentiality to honestly express themselves regarding sensitive information.
Likewise, document analysis is a useful method in a naturalistic inquiry, and it has its strengths and weaknesses. Techniques for discovery include discourse analysis and content analysis. In many cases, these kinds of documents are easily accessible, even in the case of private universities; however, not every academic library has its documents broadcast on the Internet, and some HR departments might not allow researchers access to specific job descriptions.
Weaknesses of document analysis, in this study, included access to the documents for examination; the absence of similar documents across multiple sites; and the interpretation of explicit and/or implicit content. If similar documents are not available across the academic library sites targeted for the study, the opportunity for comparison across sites is limited. Decisions on coding are subject to researcher epistemology and research goals; even if the researcher intends to focus on explicit content, it is likely that individuals do not articulate values and norms in the same way as official documents, and so some interpretation is required. This is a necessary weakness of document analysis in this case because of the interest in triangulation of methods to produce “truth” and increase trustworthiness.