The State of the Subfield

This week’s readings present some caveats as well as suggestions for the future of comparative politics research. The authors reveal that empirical analysis has been subjected to widespread misusage in the subfield. Empirical research has helped comparative politics progress in many ways, but social scientists cannot afford to lose sight of the inherent deficiencies and limitations of quantitative analysis.

A major priority on the agenda for scholars of comparative politics is that of reminding empirical researchers that their quantitative data derives from oversimplified, imperfectly measured aspects of social activity. Quantitative analysis, especially with large Ns, has steadily assumed dominance in comparative politics. While this development has helped make social theories more testable, it has been hurriedly accepted as “The” method of social science research. However, quantitativists have oftentimes lost sight of the link between numerical data and social reality. This has resulted in overconfidence, overgeneralization, and overextension of quantitative inferences.

Empirical data can be misinterpreted if: inferences are substantiated with spurious relationships of surrogate variables; observed relationships apply only under particular conditions of unmeasured variables; or, if empirical findings are too broad to explain unit-level circumstances. The last instance describes many misapplications of empirical data, such as the overlooking of long-term trends that may conceal cyclical, short-term patterns (Zuckerman 296). Misuse of empirical data can also be found in the utilization of quantitative data for explaining, rather than testing, social science theories. This caveat is described in a quotation used by Zuckerman in “Reformulating Explanatory Standards”:

These models are not ‘tested’ against historical data; rather they are
‘applied.’ The models are logical constructions, their propositions are
logically true. The question is whether they are useful, which is to say,
whether they enable the scholar to better understand the empirical world.
A model is compared to the historical record, and if it ‘fits,’ if its
propositions are reflected in the empirical record, then it is a candidate for
being an element of the explanation of that record (Fiorina 1996)(Zuckerman
295)

Empirical researchers have often made assumptions about empirical tendencies that do not reflect social phenomena. For example, non-linearity is prevalent in tendencies of social behavior, yet simple linear models are used excessively to explain nonlinear relationships (Zuckerman 297). In addition, models should anticipate probabilism, yet results are sometimes interpreted as evidence of determinism. Models should not aim to find such deterministic causal relationships, as they will inherently leave out influential variables. Models are often overextended to explain social actions, rather than probabilities of social developments. As Zuckerman explains, “Formal models provide a framework for analysis, core assumptions about the nature of political reality and the factors that drive political processes. They offer interpretations, but they do not explain particular events, decisions, or processes” (295).

The research on developing democracies and authoritarian regimes has taken a back seat in recent decades, in part due to these same empirical misusages. Third-world countries often fail to provide evidence for a model, and are accordingly left out of comparative works. The “state” has reoccupied a leading role in the subfield, which has led researchers to leave out the empirically inconvenient third-world. Universalistic conclusions are derived from these works, when in fact, they do not apply to developing states (Levi 51). This reflects a broader problem in the subfield: selection on the dependent variable. If the “state” is truly the topic of analysis, scholars ought to account for the variables that produce the divergent outcomes of developed and developing states. For the same reason, “inconvenient” autocratic states are increasingly being left out of comparative analyses, despite the dozens of enduring autocratic regimes (Geddes 367).

Theoretical universalism may be convenient for empirical research, but much research aimed at universality infers a causality chain with missing links. A series of chain links is often reduced to a single variable, leaving the fundamental components of this variable unexplored.

As a whole, the readings add weight to the claim that the subfield of comparative politics cannot advance empirically without corresponding advancements in the qualitative realm. Quantitativists cannot produce theoretically coherent findings without a thorough qualitative understanding of their data.

-Dan Blazo