On randomization-based causal inference for matched-pair factorial designs

Abstract

Under the potential outcomes framework, we introduce matched-pair factorial designs, and propose the matched-pair estimator of the factorial effects. We also calculate the randomizationbased covariance matrix of the matched-pair estimator, and provide the "Neymanian" estimator of the covariance matrix.

Improved Neymanian analysis for 2K factorial designs with binary outcomes

“…We adopt some materials from Dasgupta et al () and Lu (, ) to review the Neymanian causal inference framework for 2 K factorial designs. To maintain consistency, we inherit the set of notations from Lu ().…”

Covariate adjustment in randomization-based causal inference for2Kfactorial designs

“…, h K ) is the jth treatment combination z j . To further illustrate the construction of the model matrix, we adopt the example in Lu (2016). Example 1.…”

“…For example, randomization-based inference is applicable to the finite-population setting, and therefore may be more reasonable in practice (e.g., Miller 2006;Lu et al 2015). For more discussion on the comparison and reconciliation of randomization-based and regression-based inferences for 2 K factorial designs, see Lu (2016).…”

Randomization-based causal inference from split-plot designs

“…Despite its long tradition in the context of treatment-control experiments [Splawa- Neyman (1990), Neyman (1935), Kempthorne (1952), Imbens and Rubin (2015), Ding and Dasgupta (2016)], randomization-based inference remains an almost uncharted field when it comes to factorial experiments. The recent works of Dasgupta, Pillai and Rubin (2015), Espinosa, Dasgupta and Rubin (2016) and Lu (2016) are, to the best of our knowledge, the only literature along this line, each documenting improvements of randomization-based analysis over existing modelbased methods in the context of multi-factor completely randomized designs. Extending their methods to split-plot designs is a promising next step.…”