Developing A Theoretical Framework

We conducted several experiments and studies that contributed the qualitative and quantitative results that led to the development of the framework: AppendixBpresents several studies to explore iGesture, Chapter 3 investigates the functional utility of ges- tures, and Chapter4explores user tolerance for errors and error types for different task characteristics. Our approach to designing the framework involved using qualitative data to form the concepts that are represented in the subcategories and parameters, and quantitative data to provide the metrics used to describe the parameters. We next describe our approach to developing the framework.

Approach. Grounded theory (GT) is a qualitative research methodology that is viewed as a valuable approach to conducting research in the computing sciences (Adams et al.,2005;Elliott et al.,2002;Sarker et al.,2001). The framework was developed using an evolutionary process of subsequent applications of grounded theory. This enabled our analysis of the qualitative data which led to the categories and parameters presented in the framework. Qualitative data originated from the interviews, observations and questionnaires obtained through our experiments. This data was analysed using GT, and is discussed next.

Grounded Theory and Qualitative Research Our approach applied grounded theory to uncover the initial structure of the framework and its associated categories based on out analysis of the codes used to organise the literature review (see Appendix

A). The basic premise of using (GT) is in the systematic generation of theory from data that contains both inductive and deductive thinking, sometimes referred to as abductive reasoning (Wikipedia, 2006a; Eliasmith, 2004). While this methodology is commonly used in the social sciences for qualitative data analysis, it has also seen extensive use in the computing sciences (see Section 5.2in this chapter for references). While there are various approaches to GT, the general principal involves the generation of a theory, hypothesis or framework based on a process of repeated sampling, analysing and theorising about the data until a consistent perspective is obtained. Further analysis was applied to the qualitative results obtained through our previous experiments, which informed the current structure of the framework, shown in Figure5.1.

to Strauss, but rely on the data gathering method of Glaser’s version. There are three main components of GT, based on Strauss & Corbin (1998): Concepts, categories and propositions. Concepts are the basic elements uncovered in the open coding phase, cate- gories represent collections of concepts, and propositions are relationships or hypothesis that are made about the concepts. These are discussed next, where we outline that basic steps we used for GT.

• Open Coding - Concepts: This is first of GT, where data from field notes or transcripts are organised into concepts that are observed in the data. As more data is coded, concepts are added or merged to reflect the content. This is an iterative process of comparing, and modifying codes until the researcher has examined all the data.

• Selective Coding: After an initial round of open coding, a core concept is selected that represents a key issues that has been uncovered in the open coding. From this point, a selective process of coding around this core can guide further coding of the data. This sometimes happens after all the data is coded, however it can also be used to inform initial coding.

• Theoretical Coding - Categories: A later stage of coding, where concepts are merged together as they are compared against the rest of the data enables the researcher to uncover more theoretical concepts that emerge as a result of reviewing the bulk of the data.

• Memoing - Propositions: This is referred to as the core stage of GT methodology where the researcher writes up the ideas that have emerged through the coding process, building on the relationships and forming hypotheses that can be tested using empirical methods.

• Sorting: This stage requires structuring the data so that it can be related to others in a format that is used to connect the concepts towards a theory. While we are not at the stage of generating a theory, this part of the process represents the structure of the elements in the framework.

• Writing: This is the final stage, leading to the product of applying GT. In this case, our final product is the framework for understanding and designing gesture interactions. We present the write up in section 5.3of this Chapter, as well as a structural diagram of the framework (see Figure5.1).

While our goal is not the generation of a formal theory of gesture interactions, GT can led to the framework we present in this research. Memoing and the organisation and analysis of concepts and categories were based on hand written notes that were take for each of the participants in the studies and incorporated into the framework in several iterations. Continual applications of GT to the framework will ensure that as novel results are obtained, that the framework can be augmented and verified to address new research.

Obtaining and analysing qualitative data. In this research, we use both qual- itative and quantitative data in a cyclical approach where qualitative research informs the development of hypothesis, which is in turn tested in empirical experiments, which generates additional qualitative data. This process began with our work on the iGesture system and our analysis of the literature. This process is referred to as the qualitative, positivist research method (QPR), and is a valuable approach for conducting research in the information sciences (Straub et al., 2004). Additional references for qualitative research, grounded theory and their combination within the positivist philosophy are provided (Schloss & Smith,1999;Myers,1997;Straub et al.,2004), We next present a detailed description of the framework and its elements.