Guidelines for Conducting Mixed-methods Research: An Extension and Illustration

Abstract

In this paper, we extend the guidelines of Venkatesh et al. (2013) for mixed-methods research by identifying and integrating variations in mixed-methods research. By considering 14 properties of mixed-methods research (e.g., purposes, research questions, epistemological assumptions), our guidelines demonstrate how researchers can flexibly identify the existing variations in mixed-methods research and proceed accordingly with a study design that suits their needs. To make the guidelines actionable for various situations and issues that researchers could encounter, we develop a decision tree to map the flow and relationship among the design strategies. We also illustrate one possible type of mixed-methods research in information systems in depth and discuss how to develop and validate metainferences as the outcomes of such a study.

Understanding Topic Models in Context: A Mixed-Methods Approach to the Meaningful Analysis of Large Document Collections

“…While deductive reasoning entails arguing from the general to the particular, inductive reasoning involves the reverse scenario [49]. In particular, the type of reasoning used in a study is reflected in the assignment of codes to textual data [45].…”

“…Researchers immerse into the given data sample, and let the coding scheme emerge in a bottom-up fashion through data analysis [8,20]. Finally, in combined coding processes, scholars combine the two generic modes of reasoning stated above [28], or rely on abductive inferences along with deductive and inductive logics [49].…”

“…Mixed-methods approaches: MMR combines elements of both quantitative and qualitative research approaches within a single study [49]. Mixed methodologists take advantage of the strengths of either approach while compensating for their respective constraints [12].…”

“…Scholars increasingly draw on mixedmethods research (MMR), i.e. the combination of elements of both quantitative and qualitative research approaches within a single study [49] to examine IS phenomena [1]. Calls for methodological pluralism had already been put forth even before the IS community began to refer to this alternative as MMR [41].…”

“…Topic modeling, in particular, includes further method-specific challenges, e.g., choosing the "correct" number of topics or validating the estimated model, which are difficult to address in a purely quantitative manner [6]. The limitations of both qualitative and quantitative approaches to the analysis of large qualitative data sets have caused scholars to call for appropriate MMR alternatives [49]. Interest has been expressed in how to synergistically combine the strengths of text mining methods in general and topic models in particular with the strengths of qualitative approaches [40].…”