October 23, 2011 AMJ Special Research Forum “West Meets East: New Concepts and Theories” Idea Development Workshop
Inductive approaches • Deductive logic: Use existing theory to develop hypotheses; then collect data and test hypotheses. Most management research falls in this category. • Inductive logic: Identify a gap in the literature, collect data, analyze without a priori hypotheses to develop new theoretical constructs and relationships. • Often qualitative but doesn’t have to be – e.g., exploratory factor analyis • Often rely on interviews, but not always • Many studies use a combinationof inductive and deductive logics - i.e. “theory elaboration.” Examples from AMJ: • Elsbach and Kramer, 2003. The idea that creativity is judged on match with prototypes was already in the literature. • Gilbert, 2005. Began with several pre-defined concepts and categories. • Graebner, 2009. Used an existing definition of trust.
Inductive approaches and the SRF Inductive approaches are useful when: • No prior theory exists on an empirical phenomenon “Go beyond the Western settings to tap into the empirical phenomena of the East” • Prior theory seems unlikely to apply “Theories of work and management in organizations serving the bottom of the pyramid’ “Why” = logic that is based on Eastern context and tradition • Prior theory has gaps, especially gaps in understanding how processes unfold • “How” = relationships that differ from that of the West
Common challenges and pitfalls in inductive studies: Research design • How many cases? • Number of cases can range from 1 to 12 or more – there is no right number, just tradeoffs. • However, very difficult to develop variance theory without multiple cases. • How many interviews? • An interview does not equal a case - for a complex organizational phenomenon, one interview is not enough to create a case. • There are no firm rules on number of interviews, but if interviews are your primary data data sources, as a (very rough) guideline, it’s hard to have enough data to analyze if you have fewer than 30 interviews. • Which cases? • There are different sampling logics. Know which one you are using. • An extreme case (or polar types) is/are not the same as an idiosyncratic case.
Common challenges and pitfalls in inductive studies: Data collection • Data collection: What questions to ask? • As much as possible, focus on specific events and examples • Avoid asking general questions like “how do you manage your human resources?” • Instead, ask: “What are your hiring policies (selection process, hiring criteria? Can you give an example of the hiring process for a recently hired employee? Can you give an example of an employee who you decided not to hire?” • As much as possible, get multiple perspectives on the same incident or event • Ask the next interviewee: “Were you involved with the decision to hire (or not hire) person X? Can you tell me about that decision?” • Be sure to get some comparable data across interviewees and/or cases • Don’t limit interviews to what that informant finds most salient or interesting – you may end up with difficulty comparing across cases
Common challenges and pitfalls: Analysis • Read articles and books about analytical approaches (e.g. Miles and Huberman; Yin; Langley AMR on analyzing process data). • Read published papers as examples. • There are many ways to go about this – but be sure that you are making a conscious decision and can explain your steps in your final paper • Don’t assume you need to do formal coding or use software – these techniques lend themselves more to certain studies than others • Many qualitative, inductive researchers simply use Word tables to conduct their analysis
Common challenges and pitfalls: Writing the paper • Many authors spend too little time on the tables • For most qualitative papers, the tables are the core of your analysis and the core of your ability to persuade the reader • Once you have the tables right, writing the paper is easy – the tables are not an after-the-fact illustration • Many authors include fancy diagrams that make no sense • If you have arrows, they should all mean the same thing (typically either causality or temporal sequence) • If you have different shapes or kinds of lines, you need to explain what each type of shape or line means • Many authors fail to clearly define their constructs – don’t assume that a construct is self-explanatory.