Paper Title
Causal inference and generalization in field settings: Experimental and quasi-experimental designs
Keywords
- Causal Inference
- Generalization
- Field Settings
- Experimental Designs
- Quasi-Experimental Designs
- Rubin's Approach
- Campbell's Approach
- Causal Effects
- Regression Discontinuity Design
- Interrupted Time Series Design
- Nonequivalent Control Group Design
- Causal Inference Strengths
- Generalization of Effects
- Randomized Experiments
- Statistical Criteria
- Philosophy of Science
- Social Researchers
Article Type
Book review
Publication Info
| Pages: 40–84
Published On
March, 2000
Downloads
Abstract
This chapter introduces researchers in social psychology to designs that permit relatively strong causal inferences in the field. Considered are some basic issues in inferring causality, drawing on work by D. B. Rubin and his associates (e.g., 1974, 1978) in statistics and by D. T. Campbell and his associates (e.g., 1957) in psychology. Rubin's approach emphasized formal statistical criteria for inference; Campbell's approach emphasized concepts from philosophy of science and the practical issues confronting social researchers. These approaches are then applied to provide insights on a variety of difficult issues that arise in randomized experiments when they are conducted in the field. R. D. Cook's (1993) perspective on the generalization of causal effects is also discussed. Three classes of quasi-experimental designs are considered—regression discontinuity design, interrupted time series design, and nonequivalent control group design—that can sometimes provide a relatively strong basis for casual inference. The application of the Rubin and Campbell frameworks helps to identify strengths and weaknesses of each design. Methods of strengthening each design type with respect to causal inference and generalization of causal effects are also considered.
View more »