Thomas GreckhamerLouisiana State University

Presentation for PDWQualitative Comparative Analysis

AoM 2013, Lake Buena Vista

Building on logic of method… Illustrate logic of set-theoretic analysis Introduce QCA and its mechanics using crisp

sets (e.g. Ragin, 1987, 2000, Greckhamer, Misangyi, Elms, & Lacey, 2008)

1. Select cases & theoretically relevant attributesa) Based on substantive interest

2. Construct sets—Crisp sets distinguish ‘in’ (1, full membership), and ‘out’ (0, full non-membership)

3. Construct truth-table—all logically possible configurations of included attributes

4. Analyze set relationships between attributes and outcome of interest

5. Evaluate and interpret results

Illustrate these with simple hypothetical example

Question: What drives high firm performance in bicycle manufacturing?

Hypothetical sample of 25 manufacturers

Five key firm attributes: Firm Size Length of experience producing bicycles R & D intensity New MRP system implemented or not Vertical integration into distribution

(For empirical example, see also Greckhamer, Misangyi, Elms, & Lacey, 2008)

Define sets of firms with: Large size (alternative: e.g., “small size”) Extensive Experience High R&D intensity New MRP system Vertical forward integration High Performance

Decide membership: In crisp-sets only full membership versus full non-membership (1/0) Use theory and empirical knowledge to set

breakpoints

Common points of critique of crisp sets Information loss through dichotomization Thresholds potentially arbitrary

Some responses Maintain complexity while simplifying it When dichotomization is not straightforward,

experiment Consider fuzzy sets to enable degrees of

membership (later)

Case Experience R&Dintense NewMRP FirmSize VertInt HPFirm1 1 0 1 1 0 0Firm2 0 0 0 0 1 0Firm3 1 1 1 1 1 1Firm4 1 0 1 0 1 1Firm5 0 0 1 0 0 0

Construct Truth Table: Each row logically possible combination of crisp-

sets (2n logically possible) / Boolean expression In this example: 5 attributes = 25 = 32 logically

possible combinations

Sort cases into configurations and record outcomes & consistency

Code outcome value

Experience R&Dintense NewMRP Firmsize VertInt Number Raw Consist.0 0 0 0 0 5 00 0 1 0 0 5 0.41 1 1 1 1 5 11 0 1 0 1 3 10 0 0 0 1 2 01 0 1 1 1 2 11 1 1 1 0 2 00 0 1 0 1 1 01 0 1 1 0 1 00 0 0 1 0 0

Experience R&Dintense NewMRP Firmsize VertInt Number HP Raw Cons.0 0 0 0 0 5 0 00 0 1 0 0 5 0 0.41 1 1 1 1 5 1 11 0 1 0 1 3 1 10 0 0 0 1 2 0 01 0 1 1 1 2 1 11 1 1 1 0 2 0 00 0 1 0 1 1 0 01 0 1 1 0 1 0 00 0 0 1 0 0

Boolean algebra is used to reduce truth table to expression covering combinations with same outcome (superior performance)

Simple example: size AND R&D intensity OR size AND ~R&D intensity → superior performance size AND (R&D intensity OR ~ R&D intensity) → superior performance =size → superior performance

“Complex” solution “Parsimonious” solution “Intermediate” solution In consequent example: assume absence of new

MRP system enhances performance

Conditions Solution 1 Solution 2

Experience W WLarge Size W

R&D Intensity m

New MRP system

Vertical Integration W WConsistency 1 1Raw Coverage 0.58 0.42

Unique Coverage 0.42 0.25Solution ConsistencySolution Coverage

W W = Core/peripheral condition present

m m = Core/peripheral causal condition absent

Table 1: Configurations achieving High Performance

10.83

Do not interpret “independent effects” separate from combinations, e.g., •Experience is positively associated with high

financial performance•Size is positively associated with high financial

performance in case of experienced firms

Crisp sets contain basic logic of set-theoretic methods• Analytic unit is configuration; each is potentially

qualitatively different•Limited to 1/0 membership •Fuzzy sets are more sophisticated, rely on same

principles Systematic comparison for small N as well as

large N settings (Greckhamer, Misangyi, and Fiss, 2013)

Value of alternative model of causality