Constant Comparison

Constant comparison is the data-analytic process whereby each interpretation and finding is compared with existing findings as it emerges from the data analysis. It is associated with QUALITATIVE RESEARCH more than with QUANTITATIVE RESEARCH. It is normally associated with the grounded theory dataanalytic method, within which Glaser and Strauss (1967) referred to it as the “constant comparative method of qualitative analysis.” Qualitative and quantitative data can be subject to constant comparison, but the analysis of those data is invariably qualitative.

Each comparison is usually called an iteration and is normally associated with INDUCTIVE reasoning rather than deductive reasoning; as a result, it is also referred to as “analytic induction” (Silverman, 1993). However, hypothetico-deductive reasoning will often occur within each iteration of the constant comparison method. Constant comparison is normally associated with the IDIOGRAPHIC philosophy and approach to research rather than the nomothetic philosophy and approach. Methodologies that normally employ constant comparison include ETHNOGRAPHY, PHENOMENOLOGY, SYMBOLIC INTERACTIONISM, and ETHNOMETHODOLOGY. Constant comparison contributes to the validity of research.

An example of constant comparison might be apparent when a researcher is researching the phenomenon of leadership within an organization. The methodology might employ interview data supported by observation and document data. The initial analysis of those data might involve CODING of interview transcripts to identify the variables, or categories, that seem to be present within the manifestation of the phenomenon. After analysis of the initial interviews, a number of categories might emerge, and relationships between those categories might be indicated. With each subsequent interview, each emerging category is compared with the extant categories to determine if the emerging category is a discrete category, a property of an existing category, or representative of a category at a higher level of abstraction.

For example, Kan (2002) used the full grounded theory method to research nursing leadership within a public hospital. She determined the presence of a number of lower order categories from the constant comparison of interviews and observations tracked over a 14-month period. In addition, Kan administered a questionnaire over this time frame to measure the leadership demonstrated by the managers in question. The comparison of the questionnaire results with the categories that emerged from observation and interview secured the identification of two higher order categories called multiple realities and repressing leadership.

This ongoing, or constant, comparison continues throughout the analysis of all data until the properties of all categories are clear and the relationships between categories are clear to the researcher. In the case of the work of Kan (2002), probing into the low reliabilities of some factors and the written [Page 181]comments provided on the questionnaires provided insights that highlighted the characteristics of the higher order categories and ultimately confirmed the basic social process of “identifying paradox.”