Idiographic and Nomothetic Causal Inference in Special Education Research and Practice: Mixed Methods Perspectives

• In Event: Advancing Mixed Methods Special Education Research

Abstract

Shadish, Cook and Campbell (2002) ask whether internal validity is “the sine qua non” (p. 97) of experimental research. They explain that capacity to develop a causal argument is an essential, necessary element for experimental design research to be interpretable. In answering this question, however, they argue that molar internal validity is not the singular purpose of experimental research. One also must understand contextual interactions and the operation of mediating variables to turn the “black box” into a dynamic “clear box.” Shadish et al. go on to make the point that causal research can be studied using non-experimental methods, including case studies. This positon is further clarified by Maxwell (2005) who emphasizes that qualitative methods can form the basis for causal argumentation. Indeed, a prevailing idea is that causal inference is essentially a qualitative judgment, to be made in the context of a design and/or available evidence; statistical analyses are used to bolster related argumentation but such analyses are, by themselves, neither sufficient nor necessary for making a causal claim.
This line of thinking has bearing on special education research for at least four reasons: (1) Special education researchers often rely on group-based designs (e.g., randomized controlled trials) for obtaining causal evidence; (2) Special education researchers also utilize single-case designs to establish causal inference for individual students or classrooms; (3) There is hope that evidence from both classes of designs can be combined (cf. Hitchcock et al, 2014; Kratochwill et al., 2013) to understand intervention effects at a group difference level (i.e., nomothetic causality) and at the individual level (i.e., idiographic causality). Some treatments do not yield uniform effects across individuals, and this may be expected when working with different students with highly specialized learning needs. (4) Special education practitioners often rely on contextual/environmental information when diagnosing the causes of behavior concerns and factors that maintain behaviors, as when engaged in functional assessments (e.g., Repp & Horner, 1999). Functional assessments can entail forms of causal inference and evidence typically relies on observation, interviews, and document analyses.
These four considerations make clear that special education researchers will benefit from an overview of causal argumentation at a group level, individual student (or teacher) level, as well as distinguishing descriptive/molar causation from explanatory causation, with the latter showing mediation processes. Qualitative data have an important place in showing causal processes and mixed methods has an important place integrating these ideas into coherent wholes. A mixed methods perspective of causality (causal pluralism) is not at odds with a classic experimental design framework but rather complements it by integrating both statistical and qualitative data to frame and critique causal evidence in any number of settings and ranging from well-controlled studies to developing initial hypotheses when first meeting a student. Mixing of data can be used to yield new insights about treatment effects and causes of human behavior. This paper will therefore describe an expansive view of causal argumentation, and encourage how any threats to internal validity can be addressed by both quantitative and qualitative data.

Authors

• Burke Johnson, University of South Alabama

• John H. Hitchcock, Indiana University - Bloomington