Much of political science research is aimed at determining causality, which is defined by Johnson, Reynolds, and Mycoff as “a connection between two entities that occurs because one produces, or brings about, the other with complete or great regularity.” Essentially, causality is rooted in ascertaining whether changes in outcomes (dependent variable) are based on variance of certain factors (independent variables). Determining causality can be challenging since causation does not equal correlation and in what is called the ‘fundamental problem of causal inference’, causal inferences can never be certain due to their theoretical nature. Nevertheless, causal theories that are designed to show the causes of a phenomenon are a pivotal part of political science research and may be a key component of a thesis research paper’s hypothesis. Consequently, when designing a project, it may be useful to create diagrams in which causal mechanisms form the link or pathway from explanatory, independent variables to the outcome of interest.
Any causal claim cannot have reciprocal causation, that is, one variable must influence the other rather than multiple variables influencing each other. When studying causality, it is also important to note that time matters. The time horizons both of the cause (independent variable) and outcome (dependent variable) can impact a causal argument. Paul Pierson argues in Politics in time: history, institutions, and social analysis that each combination of different time horizons of causes and outcomes produce different categories of long term processes. For example, a long time horizon of causes and a short time horizon of outcome can create a ‘threshold effects,’ which, like an earthquake, have a minor impact until they reach a critical level at which point they cause major changes. A long time horizon of causes and a long time horizon of outcome, on the other hand, can produce ‘cumulative causes’ like global warming, in which the interaction of factors over a long period of time can have long-lasting impacts. For a full list of categories of long term processes, see Pierson, chapter 3.
In determining causation, it may be necessary to establish counterfactuals, which are hypothetical situations in which a certain causal factor is altered to be absent in order to assess its effects. For example, a researcher interested in studying the impact of incumbency on the vote share of an incumbent candidate seeking a certain office could establish a counterfactual situation in which the candidate was not an incumbent. The calculated difference between the percentage of votes the candidate would receive as an incumbent compared to as a non-incumbent would illustrate the causal effect of incumbency on a candidate’s share of votes. When formulating counterfactuals, it is important to demonstrate that the causal inference follows from both the theories and historical facts used to create it and the counterfactual proposition must be co-tenable with the counterfactual scenario. Specifically, it the counterfactual assertion had been true, then no other factors would have differed in a way that would have materially affected the outcome. For a more thorough discussion of counterfactuals, see this page of this website which is specifically dedicated to the topic.
Complex causation occurs when “an outcome results from several different combinations or conditions” (Braumoeller). For example, one could hypothesize that either the proliferation of nuclear weapons and/or the lessons learned from World War II could have led to stability during the Cold War. It is important to note that while these relationships are presented as additive, they are not in a mathematical sense, but instead are cumulative.
There are a number of different kinds of complex causal models including:
- Multiple conjunctural causation: X1 and X2 and X3 produce Y
- Substitutability: X1 or X2 or X3 produces Y
- Contexts: X2 produces Y, but only in the presence of X1
- Necessary and sufficient conditions: X1 and X2 produce Y, or X1 or X2 produces Y
- INUS (“insufficient but necessary part of a condition which in itself is unnecessary but sufficient for the result”) conditions: (X1 and X2) or (X3 and X4) produce Y
For examples of each of these types of complex causation, see Bear F. Broumoeller’s “Causal Complexity and the Study of Politics,” p.213-214.
- Bear F. Braumoeller, “Causal Complexity and the Study of Politics,” Political Analysis 11, no. 3 (August 1, 2003): 209-233.
- Gerring, John. 2005. “Causation: A Unified Framework for the Social Sciences.” Journal of Theoretical Politics 17.
- Holland, Paul W. 1986. “Statistics and Causal Inference.” Journal of the American Statistical Association 81(396): 945-960.
- Holland, Paul W. 1988. “[Employment Discrimination and Statistical Science]: Comment: Causal Mechanism or Causal Effect: Which Is Best for Statistical Science?.” Statistical Science 3(2): 186-188.
- King, Gary, Robert Owen Keohane, and Sidney Verba. 1994. Designing social inquiry: scientific inference in qualitative research. Princeton, NJ: Princeton University Press
- Lebow, R. N. 2000. “Contingency, catalysts, and international system change.” Political Science Quarterly 115(4): 591–616.
- Pierson, Paul. 2004. Politics in time: history, institutions, and social analysis. Princeton, NJ: Princeton University Press.
updated July 16, 2017 – MN