QUALITATIVE TOOLS: PROCESS TRACING AND CAUSAL PROCESS OBSERVATIONS
Process Tracing and Causal Inference
A second challenge is that of potential spuriousness: If X and Y are correlated, is this because X caused Y, or is it because some third variable caused
both X and Y? Here, process tracing can help establish whether there is a
causal chain of steps connecting X to Y, and whether there is such evidence
for other variables that may have caused both X and Y. There is no guarantee that researchers will include in their analyses the variable(s) that actually caused Y, but process tracing backward from observed outcomes to
potential causes—as well as forward from hypothesized causes to subsequent outcomes—allows researchers to uncover variables they have not
previously considered. This is similar to how a detective can work forward
from suspects and backwards from clues about a crime. It is likewise consistent with David Freedman’s argument (chap. 11, this volume) that case
expertise and substantive knowledge can play a key role in sorting out
explanations—a claim that may for some readers appear counter-intuitive
in light of Freedman’s disciplinary background as a mathematical statisticianCritics have raised two critiques of process tracing: the ‘‘infinite regress’’
problem and the ‘‘degrees of freedom’’ problem. On the former, King, Keohane, and Verba suggest that the exceedingly fine-grained level of detail
involved in process tracing can potentially lead to an infinite regress of
studying ‘‘causal steps between any two links in the chain of causal mechanisms’’ (1994: 86). Others have worried that qualitative research on a small
number of cases with a large number of variables suffers from a degrees of
freedom problem. This form of indeterminacy afflicts statistical studies,
given that the number of cases in a data set must be far greater than the
number of variables in a model to test that model through frequentist statistics
总结：
Process tracing is not a panacea for causal inference, as all methods of causal inference are potentially fallible. Researchers could fail to include an important causal variable in their analyses. Available evidence may not discriminate strongly between competing and incompatible explanations.
Actors may go to great lengths to obscure their actions and motivations when these are politically sensitive, biasing available evidence. Yet with appropriate evidence, process tracing is a powerful means of discriminating among rival explanations of historical cases even when these explanations
involve numerous variables.
On Types of Scientifific Inquiry: The Role of Qualitative Reasoning
举了一系列传染病的例子来说明定性的优势
This chapter thus seeks to demonstrate the value of causal-process observations in what could be seen as a ‘‘least-likely case,’’ that is, a data-rich
domain of mass political behavior. Even in this domain, this strategy of
causal assessment provides valuable inferential leverage that supplements,
and in this instance contradicts, the conclusions based on the analysis of
data-set observations. Indeed, the lesson for quantitative researchers is the
necessity of paying attention to the causal processes underlying behavior.
Otherwise, regression analysis is likely to jump off the rails.
Regression-Based Inference: A Case Study in Failed Causal Assessment
在【机制】方面，定性的贡献：
Scaling down to mechanisms can draw on the theoretical literature on
democracy and growth. This literature—which, after all, empirical research
on democracy and growth is designed to test—offers plausible hypotheses
about why and how democracy would cause growth, obviously involving
issues of mechanisms. For example, Olson (1983) hypothesizes that
democracy is harmful for growth because it encourages more interest
groups to seek inefficient rents from the state, reducing the efficiency of the
economy as a whole. On the other hand, Olson (1990) also proposes that
democracy may be good for growth because it encourages national leaders
to consider the economic well-being of a wider array of citizens and, consequently, to choose more efficient policies and tax rates.
表示其中内容是对原文的摘抄