You are on page 1of 15

Rationale for QCA

Method for mid-sized N studies (10-50)


Too small for statistical test
Too large for classical comp. case studies
Typical N in comparative political research
EU, OECD, German Laender, etc.
N can be much higher (individual level data)
Genuine interest in typologies
Treatment of complex causal structures
Key words: necessity, sufficiency, equifinality,
multifinality, conjunctural, asymmetry
5 November 2008 Carsten Q. Schneider 1
Rationale for QCA
Adequate for dealing with statements of
necessity and sufficiency
Set theoretically based method for set
theoretically based theories
Most social science theories are verbal and
thus set-theoretic
Aligning theories and methodologies

5 November 2008 Carsten Q. Schneider 2


Causal Complexity V
Time, Timing, Sequencing
Order of events/factors causally matters
First property rights, then economic liberalization
Functioning Market Economy (DV)
First economic liberalization, then property rights
Non-Functioning Market Economy (DV)
Antonym: static analysis
Order of events/factors on time axis is causally
irrelevant
A + B = B + A, A*B=B*A

5 November 2008 Carsten Q. Schneider 3


Necessity and Sufficiency
Powerful tools for modeling causal complexity
Frequently used but often mis-understood
Scholars mean nec./suf. but do not use the terms
Scholars use the terms but do not mean nec./suf.
Scholars confuse nec. with suf.
Often implicit in theories
Verbal statements denote set relations
Explanations of individual cases denote nec. conditions
Can be interpreted in different ways
Binary logic (crisp sets), fuzzy sets
Deterministic, probabilistic

5 November 2008 Carsten Q. Schneider 4


Necessity and Sufficiency
Whenever I see the outcome (Y), I also see the
necessary condition (X)
BUT, the inverse is not required
Whenever I see X, I also see Y
INSTEAD, cases with X but without Y do NOT
violate statement of necessity
If all cases with X also showed Y, X would be
necessary and sufficient
Formal
Values of cases in X values of cases in Y
X is superset of Y (and Y is subset of X)

5 November 2008 Carsten Q. Schneider 5


Necessity and Sufficiency
Whenever I see the sufficient condition (X), I
also see the outcome (Y)
BUT, the inverse is not required
Whenever I see Y, I also see X
INSTEAD, cases with Y but without X do NOT
violate statement of sufficiency
There might be alternative paths (sufficient condition)
for the outcome
Formal
Values of cases in Y values of cases in X
Y is superset of X (and X is subset of Y)

5 November 2008 Carsten Q. Schneider 6


2x2 Table: Example
Hypothesis: Inhabitants Not
of small villages in the Cons. directly Many
Alps (IV) are voter relevant cases
conservative voters
(DV) Not
Expected data pattern: Non- directly Few
set relation is in line cons. relevant cases
X is subset of Y voter
X is sufficient for Y Not Alpine
Alpine Inhabita
inhabita nt
nt
5 November 2008 Carsten Q. Schneider 7
2x2 Table: Example
Hypothesis: Inhabitants
of small villages in the Cons. Few Many
Alps (IV) are voter cases cases
conservative voters
(DV)
correlation is Non- Many Few
NOT in line with cons. cases cases
hypothesis voter
Not Alpine
Alpine Inhabita
inhabita nt
nt
5 November 2008 Carsten Q. Schneider 8
Truth Tables
Concept stems from formal logic
No ‘truth’ in a higher (moral, theological etc.)
sense
At the core of any QCA analysis
QCA as a research approach
Creation of truth table
Often takes years
QCA as a data analysis technique
Analysis of info contained in truth table
Often takes just some days

5 November 2008 Carsten Q. Schneider 9


Truth Tables
How do we get from a data matrix to a truth
table?
Write down all 2k combinations
Look at each case and check which
combinations of conditions it displays and put
it into the respective truth table row
Check which outcome value the case displays
and assign that value for the truth table row in
the outcome column

5 November 2008 Carsten Q. Schneider 10


Truth Tables vs. Data Matrix
Conditions Outcome

Cases A B C Y
ARG 1 1 1 0
PER 1 0 0 0
BOL 1 1 0 0
CHI 0 1 0 1 DATA MATRIX
ECU 1 0 0 0
Y = stable democracy
BRZ 0 1 1 1 A = violent breakdown
B = ethnic homogeneity
URU 1 0 1 1
C = fragmented party system
PAR 0 0 1 1
COL 0 0 0 1
VEN 1 1 1 0
5 November 2008 Carsten Q. Schneider 11
Truth Tables (V)
Conditions Outcome

Row A B C Y
1 0 0 0 1
2 0 0 1 1
3 0 1 0 1 TRUTH TABLE

4 0 1 1 1 Y = stable democracy
A = violent breakdown
5 1 0 0 0
B = ethnic homogeneity
6 1 0 1 1 C = fragmented party system
7 1 1 0 0
8 1 1 1 0
5 November 2008 Carsten Q. Schneider 12
Analysis of Truth Tables
How to analyze a truth table?
Aim
Which (combinations of conditions) are linked to the
outcome?
Which combinations are sufficient for the outcome?
Answer
Any row that displays the outcome
Most complex answer/solution term
How to get to more parsimonious solution?
By hand
Computer

5 November 2008 Carsten Q. Schneider 13


Analysis of Truth Tables (II)
Procedure of minimizing truth table
Write down info in Boolean notation
row 1 + 2 + 3 + 4 + 6

abc + abC + aBc + aBC + AbC Y


row 5 + 7 + 8
Abc + ABc + ABC y
Represents causal complexity
Equifinal - Logical OR (+)
Conjunctural - Logical AND (*)
Asymmetric - different solutions for Y and y
Multifinal - e.g. condition C

5 November 2008 Carsten Q. Schneider 14


Analysis of Truth Tables (III)
abc + abC + aBc + aBC + AbC Y

ab + aB + bC Y

a + bC Y

Primitive expression – Prime Implicants – Minimal


Solution
contain same truth value contained in truth table
are logically equivalent
should be reported in publications
Which one to focus on most depends on research aims
5 November 2008 Carsten Q. Schneider 15

You might also like