Process-Tracing Methods: Foundations and Guidelines

Glossary

175 Glossary Abduction: A method or approach that is a dialectic combination of deduction and induction (see Peirce 1955). Account evidence: A type of evidence in the form of a statement or narrative that gives a description of an event or thing. Common sources of account evidence include participant interviews and documents (e.g., minutes of meetings). Accuracy of obser vations: Whether we are measuring what we intended to measure . Inaccurate measures can be the product of either nonsystematic or systematic errors in the measuring instrument we use to collect observations. Nonsystematic (or random) error is commonly termed reliability, whereas systematic error is defined as the bias in our measuring instrument. Accuracy in Bayesian terms can be represented as the probability or degree of confidence we have that a measure is accurate, depicted as p(a). Activity: Part of a causal mechanism. Each part of a causal mechanism is composed of entities that engage in activities that transmit causal forces through a causal mechanism. Activities are the producers of change, or what transmits causal forces through a mechanism. Activities are conceptualized as verbs, denoting activity (Machamer 2004; Machamer, Darden, and Craver 2000). Note that nonactivity can also be conceptualized as an activity. Bayesian logic of infer ence: A logical formula for estimating the probability that a theory or hypothesis is supported by found evidence based on the researcher’s degree of belief about the probability of the theory/hypothesis and the probability of finding given evidence if the theory or hypothesis is valid before gathering the data. Case-centric research: Research that is more interested in explaining the outcome of a particular case than a generalizing ambition to make theoretical claims beyond the single case. Outcomes are defined in a broader manner, including all of the important aspects of what happened in a case such as the Cuban Missle Crisis. See also Systematic mechanisms; Nonsystematic mechanisms. Causal effect: The difference between the systematic component of observations made when the independent variable takes one value and the systematic component of comparable observations when the independent variable takes on another value (King, Keohane, and Verba 1994: 81–82). 176 Glossary Causal inference: The use of observed empirical material to make conclusions about causation, understood as either patterns of regularity (mean causal effects) or mechanismic relations. Causal mechanism: Theorized system that produces outcomes through the interaction of a series of parts that transmit causal forces from X to Y. Each part of a mechanism is an individually insufficient but necessary factor in a whole mechanism , which together produces Y. The parts of causal mechanisms are composed of entities engaging in activities. Causal process observation (CPO): A term used to describe “pieces of data that provide information about context, process, or mechanism and contribute distinctive leverage in causal inference” (Seawright and Collier 2010: 318). The term CPO overlaps to some extent with our use of the term evidence. Causality in the mechanismic sense: An ontological understanding where causation is confirmed only when an underlying mechanism can be shown to causally connect X and Y. The mechanismic understanding of causality does not necessarily imply regular association. Causality as regularity: An ontological understanding of causality where causation is defined as a pattern of constant conjunction between X and Y. For causality to be established, Hume argued that three criteria for the relationship between X and Y needed to be fulfilled: (1) X and Y are contiguous in space and time; (2) X occurs before Y (temporal succession); and (3) there is a regular conjunction between X and Y (Holland 1986). Certain predictions: Predictions about evidence that are unequivocal, where the prediction must be observed or else the empirical test disconfirms the existence of the part of the mechanism. Also termed disconfirmatory power. Classical statistical methods: Methods that assess the mean causal effects of X on Y in a population. Inferences are made using traditional statistical theory (e.g., Fisher, Pearson) and using tools such as the Central Limit Theorem (termed the frequentist logic of inference in this book). The empirical material used is large numbers of data-set observations. Only cross-case inferences to the population of a phenomenon are made. Also termed large-n statistical methods. Comparative cross-case methods: Methods used to assess necessary and/or sufficient conditions that produce Y. Involves using different tools, such as Mill’s methods of agreement and disagreement. The methods enable cross-case inferences to be made to for a population of cases, although the scope of populations...