Abstract
Comparative syllogism is a type of scientific reasoning widely used, explicitly or implicitly, for inferences from observations to conclusions about effectiveness, but its philosophical significance has not been fully elaborated or appreciated. In its simplest form, the comparative syllogism derives a conclusion about the effectiveness of a factor (e.g. a treatment or an exposure) on a certain property via an experiment design using a test (experimental) group and a comparison (control) group. Our objective is to show that the comparative syllogism can be understood as encoding a simulation view of counterfactuals, in that counterfactual situations are conceptual constructs that can be correctly simulated by homogeneous comparison groups. In this simulation view, the empirical data from the comparison groups play an evidential role in the evaluation of counterfactuals and in obtaining counterfactual knowledge. We further indicate how successful experimental designs can help us to obtain correct simulations, and thus to bring us to scientifically-empirically based counterfactual knowledge.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Vandenbroucke (2002, p. 261) indicates that the use of comparison groups was already recorded in the Old Testament, so the technique has a long history.
See, among others, Morgan and Winship (2007) for a comprehensive survey and references on the theory of causal estimation.
See, for example, Howson and Urbach (2006, Chaps. 6 and 8).
It is controversial exactly how to indicate counterfactuality, but here we fall back on the strategy of using subjunctive mood.
In the literature, the counterfactual measure is also called the potential outcome. The counterfactual measure as a pure conceptual construct is also labeled as being metaphysical (cf. Morgan and Winship 2007, Chap. 10).
In the scientific practice, people ususally do not take the test group (with a treatment) to simulate the counterfactual measure of the comparion group (without the treatment), since the main practical concern is on the effect of the actually excuted treatment which makes the situtation of not having the treatment counterfactual. In Sect. 6, we explore the inference pattern that uses the test group to simulate the counterfactual measure of the comparison group.
In the literature, other “terms” are used to express that confounding is a conceptual construct. For example, Vandenbroucke (2002, p. 221) indicates that confounding is an a priori notion, and indicates that the notion of confounding was recognized as an a priori notion when it was developed in the department of epidemiology at the Harvard School of Public Health in the 1970s. Rothman et al. (2008, p. 57) indicates that the notion of confounding is “metaphysical” in order to express this “a priori” concern.
In a sense, this is similar to Quine’s introducing abstract sets into scientific ontology.
This homogeneity requirement is intuitively similar to requiring that the comparison group is, ceteris paribus, identical to the test group.
This is to say that \(C_{f}\) is a risk factor for Y in the sense of epidemiology.
In epidemiology, the group A is in general called the exposed group, and the group B is called the unexposed group.
In epistemology, for example, X can be an exposure and Y a disease.
A similar point is made in Woodward (2003, pp. 104–107) for an interventionist notion of causation, when he tries to define causation by intervention, though at the same time causation is also involved in the definition of intervention.
Hall (2007, p. 110) similarly indicates that if defining a counterfactual measure depends on the counterfactual measures of other factors, then the defined counterfactual measure represents only aspects of other antecedently understood counterfactual measures. A similar worry about the evaluation problem also arises for theories of counterfactuals that make use of the cotenability condition but at the same time define the cotenability condition in terms of counterfactuals (cf. among others, Goodman 1955, p. 16, Barker 1999, pp. 436–437).
In the literature, there are two ways to de-confound (control confounding): one is to de-confound by experimental designs, and the other uses data analysis (cf. Greenland et al. 1999). The former is directly related to providing a better ground for the comparative syllogism, but the latter instead gives a different way to analyze data in order to avoid arriving at a confounded conclusion. For more on these and other methods of de-confounding, see also Rothman et al. (2008, Chap. 9).
Pearl (2000, p. 184) suggests that \(m_{t}^{0}\) may be interpreted as the measure of a “randomized comparison group.” In this paper, we focus on randomization as an experimental method to find comparison groups to simulate \(m_{t}^{0}\), based on defining \(m_{t}^{0}\) as \(m_{h}^{0}\).
See also, among others, Jager et al. (2008).
See also Hiddleston (2005) for an appeal to the minimality condition.
References
Barker, S. (1999). Counterfactuals, probabilistic counterfactuals and causation. Mind, 108, 427–469.
Dawid, A. P. (2000). Causal reasoning without counterfactuals. Journal of the American Statistic Association, 95, 407–424.
Dawid, A. P. (2002). Commentary: Counterfactuals: Help or hindrance? International Jounral of Epidemiology, 31, 429–430.
Goldman, A. (1976). Perceptual knowledge and discrimination. Journal of Philosophy, 73, 771–791.
Goodman, N. (1955). Fact, fiction and forecast. Cambridge: Harvard University Press.
Greenland, S., & Morgenstern, H. (2001). Confounding in health research. The Annual Review of Public Health, 22, 189–212.
Greenland, S., & Robins, J. (1986). Identifiability, exchangeability, and epidemiological confounding. International Journal of Epidemiology, 15, 413–419.
Greenland, S., & Robins, J. (2009). Identifiability, exchangeability and confounding revisited. Epidemiologic Perspectives & Innovations, 6, 4.
Greenland, S., Robins, J., & Pearl, J. (1999). Confounding and collapsibility in casual inference. Statistical Science, 14, 29–46.
Hall, N. (2007). Structural equation and causation. Philosophical Studies, 132, 109–136.
Hiddleston, E. (2005). A causal theory of counterfactuals. Nous, 39, 632–657.
Hill, C. (2006). Modality, modal epistemology, and the metaphysics of consciousness. In S. Nichols (Ed.), The architecture of imagination (pp. 205–236). Oxford: Oxford University Press.
Hitchcock, C. (2001). The intransitivity of causation revealed in equations and graphs. Journal of Philosophy, 98, 273–299.
Hitchcock, C. (2007). Prevention, preemption, and the principle of sufficient reason. Philosophical Review, 116, 485–532.
Howson, C., & Urbach, P. (2006). Scientific reasoning: the Bayesian approach (3rd ed.). La Salle, IL: Open Court.
Jager, K., Zoccali, C., MacLeod, A., & Dekker, F. (2008). Confounding: What it is and how to deal with it. Kidney International, 73, 256–260.
Jenkins, C. S. (2008). Modal knowledge, counterfactual knowledge and the role of experience. Philosophical Quarterly, 58, 693–701.
Kvart, I. (1986). A theory of counterfactuals. Indianapolis, IN: Hackett.
Lewis, D. (1973). Counterfactuals. Oxford: Blackwell.
Maldonado, G., & Greenland, S. (2002). Estimating causal effects. International Journal of Epidemiology, 31, 422–429.
McNamee, R. (2003). Confounding and confounders. Occupational Environmental Medicine, 60, 227–234.
Morgan, S., & Winship, C. (2007). Counterfactuals and causal inference: Methods and principles for social research. New York: Cambridge University Press.
Nozick, R. (1981). Philosophical explanation. Cambridge: Harvard University Press.
Pearl, J. (2000). Causality: Models, reasoning, and inference. Cambridge: Cambridge University Press.
Pritchard, D. (2005). Epistemic luck. Oxford: Clarendon Press.
Rosner, B. (2010). Fundamentals of biostatistics (7th ed.). Pacific Grove, CA: Duxbury Press.
Rothman, K., Greenland, S., & Lash, T. (2008). Modern epidemiology (3rd ed.). Philadelphia: Lippincott Williams & Wilkins.
Sauchelli, A. (2010). Concrete possible worlds and counterfactual conditionals: Lewis versus Williamson on modal knowledge. Synthese, 176, 345–359.
Stalnaker, R. (1968). A theory of conditionals. In Studies in logical theory, American philosophical quarterly monograph series 2 (pp. 98–112). Oxford: Blackwell.
Steel, D. (2007). Across the boundaries: Extrapolation in biology and social science. New York: Oxford University Press.
Vandenbroucke, J. (2002). The history of confounding. Sozial und Praventivmedizin, 47, 216–224.
Wang, L., & Ma, W.-F. (2012). Scientific knowledge and extended epistemic virtues. Erkenntnis, 77, 273–295.
Williamson, T. (2007). The philosophy of philosophy. Oxford: Blackwell Publishing.
Woodward, J. (2003). Making things happen. New York: Oxford University Press.
Worrall, J. (2002). What evidence in evidence-based medicine? Philosophy of Science, 69, 316–330.
Worrall, J. (2007). Why there’s no cause to randomize. The British Journal for the Philosophy of Science, 58, 451–488.
Acknowledgments
We would like to thank three anonymous reviewers for very helpful comments. Funding for this study was supported by research grants of Taiwan National Science Council (NSC-100-2410-H-194-085-MY3, NSC-101-2511-S-039-003).
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Wang, L., Ma, WF. Comparative syllogism and counterfactual knowledge. Synthese 191, 1327–1348 (2014). https://doi.org/10.1007/s11229-013-0330-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11229-013-0330-0