1
A Boolean Approach To Party Preference. A Five-Country Study.
ABSTRACT
Research problem. Qualitative Comparative Analysis (QCA) overlaps logistic regression in
explaining events but challenges the latter's lack of accounting for causal complexity. QCA
has only to a limited degree been applied to individuals as cases and has not incorporated the
logic of probability. Objective. QCA and logistic regression are compared with respect to
logic, procedure and outcome in explaining individuals' party preferences. Data. Constructs of
types of equality from surveys in five Nordic countries are adapted to the requirements of the
two methods. Results. QCA and logistic regression converge and overlap in identifying
degrees of causal complexity, to ascertaining model significance, and to identifying key
causal antecedents to party preference. Conclusion. Results differ in degree, not in kind. A
slightly more nuanced picture emerges using the QCA approach whereas logistic regression
delivers greater parsimony. The issue of causal complexity relates primarily to theory and
design and secondarily to method. Social research needs diverse tools.
1. INTRODUCTION
Qualitative comparative analysis (QCA) and logistic regression converge on explaning events.
They diverge on two broad counts. First, the explicit assumption of analysts is that QCA,
based on Boolean algebra, in accounting for causal complexity, is superior to other methods
which, in turn, fall short of accounting for important behavioural relationships. The QCA
point of departure is twofold. First, it assumes maximal complexity. And second, it assumes
that an exhaustive combination of all conditions is the most meaningful way to approach and
identify causal patterns. If empirical complexity exists, a larger set of causal conditions will
be identified. If empirical simplicity exists, a smaller set of causal conditions will still prevail.
Logistic regression adds, multiplies and averages selected independent variables in order to
measure effects. These effects are controlled against each other and sanctified through
statistical significance. The implicit assumption of logistic regression is to offer parsimony
and power. Given the methods' different points of departure one may expect, when using the
same data, a divergence between the two methods in their accounting for degree of causal
complexity. And from the methods' different procedures one may expect somewhat diverging
Convention