|
|
|
|
|
Sorting Out Card Sorting |
|
|
|
|
|
[Home] [Abstract] [TOC] [Chapter 1] [Chapter 2] [Chapter 3] [Chapter 4] [Chapter 5] [Appendix A] [References] |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Summative Reflections of the Researcher The complexities of Usability, Human Computer Interaction, and Information Architecture are both greatly removed from and intricately intertwined with the complexities of electronic information systems. Over the years, as a veteran of the technical end of electronic information systems I have become frustrated with supporting systems that were designed according to the interests and perspectives of the technology experts, rather than the interests and perspectives of users or needs of an organization. After years of cacophonous complaints about various system designs, I happened across a brief description of card sorting, presented as a “deceptively simple” (Boorman & Arabie, 1972) method of gaining insight into user preferences for the design of an information system. My reaction was, what could be easier or more intuitive than writing labels that represent content or tasks on a stack of recipe cards and asking users to sort them any way they saw fit? It brought to mind a favorite quote of Albert Einstein, who aptly stated, “Any intelligent fool can make things bigger and more complex …it takes a touch of genius – and a lot of courage to move in the opposite direction” (Einstein, n.d.). With no prior experience in card sorting beyond a ten-minute exercise in a University of Minnesota workshop and a brief assignment using the online tool CardZort® in a graduate Taxonomy course at the University of Oregon, I approached this study with limited knowledge, experience, and no preconceived notion of how to design, conduct, or analyze a card sort. The primary intent of this research is to provide practitioners with a global overview of card sorting methods as described in the literature under review. This overview is presented as a table of criteria for practitioners to consider when designing a card sorting exercise. Discussion of the criteria presented in Chapter IV - Analysis of Data should not be interpreted as either conclusive or inclusive. As a preliminary study, considerable potential exists for the further identification of categories and for the addition of properties to existing categories. Thus, the reader is highly encouraged to study these resources to make their own determination on the validity of constructs used by this researcher for the classification and assignment of properties to the characteristics. Observations on Grounded Theory and the Constant Comparative Method The constant comparison method (Glaser and Strauss, 1967) demonstrates considerable advantages when applied to this type of study. Chapter IV – Analysis of Data represents the author’s interpretation of the data and is “concerned with generating and plausibly suggesting (but not provisionally testing) many categories, properties, and hypotheses about general problems” (Glaser & Strauss, 1967, p. 104). This study proposes a method to categorize the characteristics of a card sorting method and to identify the properties of those categories. No attempt is made to test the reliability or validity of the card sorting methods. In Chapter III – Method the researcher states that “this study does not draw conclusions from the data”. The reasoning behind this claim is that grounded theory research seeks to discover themes and patterns that emerge from unstructured verbal or written data – interviews, books, literature, field notes, observations, and other sources (Glaser & Strauss, 1967). The categorization and assignment of properties to a phenomenon may lead the researcher to the formulation of a theory – a hypothesis - a research question. Grounded theory research does not seek test a hypothesis (Glaser & Strauss, 1967). Thus, the conclusions of this research are represented by the outcome of this study - a prototype tool – a method – a decision support system intended to assist practitioners with the design of a card sorting exercise. This study does not propose what that design should be. Speculations on the Tool The capability of the hypertext table (see Table 1: Twelve Categories of Card Sorting Characteristics) may be approaching practical limits for presenting the data gleaned from the hybrid conceptual analysis and constant comparative process. As further categories and properties are defined and additional literature or practitioner data are added, the need to scroll up and down or left and right becomes cumbersome and frustrating, making cognitive absorption problematic. Because of the limitations inherent in the spreadsheet table, a proposal is made for the development of a type of expert system (an extensible database) designed for comparing card sorting methods. In this system, qualified practitioners would complete an extensive online survey on card sorting methods, assigning Likert scale ratings to selected properties of the categories and elaborating with free text comments that semantically reinforce their concepts and ratings. By assigning tangible properties to categories of quantitative and qualitative card sorting characteristics, structured queries could filter the data. The results of the query would be presented in a condensed form on a web page or as a downloadable table or document. Practitioners could then review the filtered data to assist them with the design of a card sorting exercise “based on the collective wisdom of the industry-wide community of UCD [user centered design] practitioners” (Carey, et al. 2002). Conclusion Card sorting is a research method, often applied in one form or another in the social sciences, that seeks to increase our understanding of human thought and behavior. Coxon (1999) contends that sorting and categorization is “the most fundamental operation of thinking and language” (Coxon, 1999). In many ways, this study evolved in a self-reflexive manner – one intrinsically about itself - because it applied a social science research methodology to the analysis of a research methodology used in social science. Using the cognitive power of sorting to further our understanding of sorting may help us gain insight into our own mental models. Although quantitative statistical methods may be necessary to aggregate results from card sorts with a large number of participants and cards, the transformation of any card sorting analysis into a final design requires a measure of insight and intuition, a process described by Mauer and Warfel (2004) as “part science, part magic” (Mauer & Warfel, 2004). Researchers should continually strive to objectively improve research design and analysis methods, keeping in mind the thoughts of Deaton (2002) who concludes, “Regardless of how you analyze your data, the design decisions you reach should still be guided by your experience as a design professional” (Deaton, 2002). |
|
|
|
|
|
Read in Browser | Download PDF ©2005
Steven D. Hannah
|