4.35 Design-based Approach

  • “a mode of statistical analysis that emphasizes design rather than statistical modeling” (Keele 2015b, 12)
    • Design: “all contemplating, collecting, organizing, and analyzing of data that takes place prior to seeing any outcome data” (Rubin2008, 810; Keele 2015b, 12)
      • Q: Does that include measurement? YES!
  • “[S]et of techniques that can make identification more credible without the use of parametric statistical models and without using outcomes.” (Keele 2015b, 2)
  • Given an identification strategy (e.g. natural experiment) “analyst can often use elements of the design-based approach to improve the research design.” (Keele 2015b, 2)
  • Techniques: Reduce heterogeneity (Q: Norvell and Cummings 2002 on effects of helmets?); Falsification tests [search treatment effect where it should not exist]; Sensitivity analysis etc.
  • Design-based causal analysis seeks plausible identification strategy and employs techniques above to bolster the credibility of that strategy

References

Keele, Luke. 2015b. “The Statistics of Causal Inference: A View from Political Methodology.” Polit. Anal. 23 (3): 313–35.

Norvell, Daniel C, and Peter Cummings. 2002. “Association of Helmet Use with Death in Motorcycle Crashes: A Matched-Pair Cohort Study.” Am. J. Epidemiol. 156 (5): 483–87.