Design of Qualitative Research
Research questions appropriate for qualitative analysis. Qualitative Research: Grounded Theory Approach Construct-related questions Nature of new constructs Typologies Mechanisms & Dynamics Process Models (Mohr basis for path model) Playing out of change Variance Models Capture nature of relationships Linear Curvilinear Capture complex interactions Capture antecedents and consequences of deep /hard to measure constructs Meaning Identity
Theory-Driven Research: Two Approaches, Two Objectives Approaches Deductive: Generate theory on the basis of logic and a priori assumptions (armchair theorizing). Inductive (Grounded): Generate theory by discovery from the data. (Exampling = middle ground; cases illustrate & tweak deduced propositions) Goals Generation of theory Verification of theory
Qualitative vs Quantitative Research: When & How Applied? Generate Verify Qualitative Grounded theory Content analysis Quantitative Phenomena-driven; epidemiological research Mainstream theory testing (assumes that we have enough theory, but insufficient verification)
Grounded theory method Focus on generating theoretical ideas (or hypotheses) from the data rather than having these ideas specified beforehand A grounded theory is one that is inductively derived from the study of the phenomena it represents. Strauss and Corbin. p.23 Key focus = reflective reading of text and the application of codes
Origins Developed in the School of Nursing, University of California San Francisco by sociologists Barney Glaser and Anselm Strauss in their book: Awareness of Dying Glaser Strauss
Philosophical/disciplinary contexts Social Science Paradigm in 1960s Quantitative, hypothesis testing, surveys Lazarsfeld Symbolic Interactionism Behaviour shaped by the interactive construction of meaning. Phenomenology Need to set aside preconceptions Emphasizes the lifeworlds of the actors. Focus on rich (albeit often mundane) detail of lived experiences. Describe; focus on language; stay authentic; don t over-interpret Ethnography can be phenomenological (strictly descriptive) or in the service of theory.
Core Elements of Grounded Theory Inquiry shaped by the aim to discover social & social psychological processes. Create analytic codes and categories from the data Data collection and analysis phases of project proceed simultaneously. Analytic process employed prompts theory discovery and development rather than verification of preexisting theories = Inductive Theoretical sampling refines, elaborates and exhausts conceptual categories. Systematic application of grounded theory analytic methods will progressively lead to more abstract analytic levels. (Charmaz, 1983 p. 125)
Grounded theory (Glaser & Strauss 1967; Strauss & Corbin 1990) Data Theory Theory Data - Concepts and relations between them come from the data and are worked out in relation to the data during the course of the research (Glaser & Strauss) - Formulating theoretical interpretations of data grounded in reality provides a powerful means both for understanding the world out there and for developing action strategies that will allow for some measure of control over it (Strauss & Corbin, 1990, p. 9)
Principles of Comparative Analysis (1) Theory generation: Identify conceptual categories & propose links between them Category identification is based on comparison. We do the following relative to other constructs: Original categories; don t borrow existing ones. Show evidence of defining properties Show evidence of implicit discriminant validity Show evidence of category stability (properties don t change, or if they do, may indicate new category?) Show evidence of its generalizability (or specificity in time or place).
Principles of Comparative Analysis (2) Develop a conceptual hierarchy (tree). Must be logical yet sensitizing (give examples) Categories have sub-categories which may be broken down into further sub-sub-categories. All on the basis of conceptual properties Generates model of conceptual structure Stressor Role Conflict Role Overload Inter Intra Qualitative Quantitative
Principles of Comparative Analysis (3) Linkages between categories are identified by looking for comparative patterns. Which patterns of relatedness are most dominant? What conditions determine pattern existence or strength? Search for pattern (and pattern variation) across categories of cases. Suggest, rather than prove. One case sufficient to identify category or pattern of links. Additional cases needed to assess generalizability and conditionality.
Principles of Comparative Analysis (4) Simultaneous data collection, coding and analysis. Look for linkages as sub-categories are identified and properties defined. Iterative process Comparative analysis based on data already collected but also drives next round of data collection. Evidence must be both confirmatory and disconfirmatory.
Typologies Emic indignant to the group studied; used by informants Example: Giallombardo s study of WOMEX women s prison. Snitchers, squares, jive bitches, hustlers, boosters, inners, cherries, punks and turnabouts Etic imposed by the outsider (researcher): Example: Lofland s study of waiting behavior The sweet young thing, investigator, maverick, etc. Used extensively in market research
Variance vs. Process Theories Type of concepts Variance Entities that have varying values Process Entities that participate in or are affected by events Change in concepts over time Properties do not change over time (only their values change) Entities can change over time Types of relationships Variation among values of properties Sequences among events (typically probabilistic) Time ordering in the relationships among concepts Causal logic in the relationships among concepts Time ordering among independent variables (properties) is immaterial Causal logic is desired, but association is critical Time ordering of events is important Causal logic is critical
Process Models
Rules of Evidence Generating Identify unique concept categories Conceptual categories can emerge from one case or situation. Properties must be defined. Verifying - Look for similarities and differences across cases/situations Compare same category in alternative cases or situations. Look for situation-specific properties. If situation-specific properties found, there is evidence of sub-categories. Disconfirmation. Similar generating/verifying process for casespecific relationships. Objective: Develop a theory that accounts for much of the relevant behavior.
Theoretical Sampling: General Joint collection, coding and analysis of data with aim of determining what additional data are needed and from whom, where and when in order to maximize comparisons. Data needs are determined as theory emerges Initial sample is determined on the basis of early theoretical speculation. Initial concepts drive determination of who/what to sample. Sample is expanded by focus on disconfirmation. Where is my categorical framework least likely to hold? Among whom are the linkages discovered likely to be disconfirmed?
Theoretical Sampling: Choosing Comparative Groups Choose groups based on their theoretical relevance for emerging categories and linkages. Determination of who to retain: Set by boundary conditions of theory. If study of older workers, then exclude those < age 52. Within boundary conditions, sample especially those: Who are the least comparable to mainstream highlights differences. Who are the most comparable to mainstream highlight similarities. Make conscientious effort to identify types of individuals, groups, organizations associated with phenomena under exploration. Sample several from each type Seek out differences (sub-types) within types and/or groups within a type for who the category or relationship is different.
Theoretical Sampling: WHAT/WHEN & HOW MANY to Sample WHAT: Interviews, observations, archival data (Triangulation) Rules of evidence: If one source offers constrained view, need to collect data from additional sources. Expand sources until alternative courses no longer expand categories or category-specific properties. HOW MANY: Depth of data collection No need to collect same information from all informants. Target specific informants for specific data. As collection continues, depth is substituted for breadth Identify what additional information is sought from which informants. Use targeted informants to attempt to disconfirm WHEN: Timing Process theories demand sampling by time/phases Fewer informants, but many within/person data points. Variance theories Do categories and category-specific properties remain stable over time? Do relationships vary over time? Process-dependent? Sample to map pathways over time
Theoretical Sampling: Saturation Sampling continues until investigation of additional groups: fails to generate additional categories or categorical properties. Despite group differences, emergent properties are similar/identical. Minimizing differences across comparison groups: Establishes conditions for a category to exist. Offers evidence of discriminant validity. Provides basis for predictive validity.
Example from Boundaries in Helping Study Initial Categories Helping for substance misuse cases Helping for interpersonal problems at work Helping for task-related problems at work Helping for family matters & matters outside of work Groups sampled Peer Counselors Within case category above, sampled by: Male vs. Female Old vs. Young Occupation Supervisor working with peer counselor on each type of case above, sampled by: Supportive Unsupportive Union leader working with peer counselor on each type of case above, sampled by: Supportive Unsupportive
Theoretical vs. Statistical Sampling Theoretical Sampling Objective to discover categories and their properties Uncover likely relationships among categories Evaluate sample on the basis of: Persistence of unexplained exceptions (determines sample size) Exclusion of boundary-relevant groups (e.g., all interviewees are male). Statistical Sampling Validate conceptual structures Demonstrate support for proposed linkages Estimate predictive utility or explanatory potential Evaluate sample on the basis of: Generalizability Statistical power (determines sample size) Internal bias
Categories and the Constant Comparative Method Content analysis vs. Constant Comparison: Both are Qualitative! Content analysis: Preconceived categories Objective: Hypothesis testing Tool: Statistical analysis Focus is on: Distribution of cases among these categories. Pattern of variation and co-variation across these categories according to some underlying logic (over-arching theory).
Content Analysis Example CEO Narcissism and Extreme Returns on ROA Theory: Narcissistic CEOs are over confident & swing for the fences Study Design (Random assignment? Survey?) Annual reports, press releases, interviews, comp N. Index Photo size Pronoun use (first-person singular [ I ] vs. plural [ we ] Pay differentials Correlation btwn photo size & pay gap =.49 Security analysts survey Correlation Analysts score & N. Index? 0.82 Correl. N. Index & Extreme ROI? 0.17 Source: Chatterjee & Hambrick, ASQ, 2007
Categories and the Constant Comparative Method Content analysis vs. Constant Comparison: Both are Qualitative! Content analysis: Preconceived categories Objective: Hypothesis testing Tool: Statistical analysis Focus is on: Distribution of cases among these categories. Pattern of variation and co-variation across these categories according to some underlying logic (over-arching theory). Constant Comparison: Emergent categories Objective: Construct/Theory generation Tool: Memos and additional data collection as data are analyzed Focus is on: Discovering categories and properties Discovering patterns of variation and co-variation Using data to suggest the logic underlying observed patterns.
Four Stages of Constant Comparison 1. Comparing incidents applicable to each category Code incidents into as many categories as possible. Provide a rule for the code: What properties are necessary vs. sufficient Compare incident to other incidents & note discriminating properties. Write memos to highlight: Dilemmas in coding (suggest discriminant validity) Repeating patterns of covariation in codes (suggest links) 2. Integrating categories and their properties Shift from comparing incidents to comparing incidents to categories Identify relations among categories (i.e., trees) Identify conditioning properties (e.g., demographics, temporal phases, roles) determining branches and subcategories 3. Delimiting the theory Coding results in fewer modifications of properties (saturation) Core categories and category-based relations emerge Code for depth of understanding of these relationships and conditions governing them. Parsimony in constructs; Reduction in theoretical scope 4. Writing the theory Memos provide content to explain categories (content, construct & discriminant validity) Memos provide logic governing emergent patterns (conditions governing relations suggest mechanisms)