Theoretical rationale for privileging pretests in bias control: high correlation with both outcome and selection process¶

Theory · All theories
Evidence · no claims cited

Description¶

The article offers an explanatory account of why pretest measures of the study outcome are privileged tools for controlling selection bias in observational studies. The rationale rests on two correlations: pretests are "often highly correlated with the outcome," and "in many contexts, they are also highly correlated with the selection process," so measuring them should capture much of both the outcome signal and the selection into treatment. The article tests this rationale empirically and finds the resulting bias-elimination pattern more complex than the rationale predicts.

Design Implications¶

Context¶

Requirements¶

• Pretest measures must be of the same outcome as the study outcome for the rationale to apply as stated

Constraints¶

• The article's own findings show bias elimination does not follow the predicted pattern based on the pretest's correlation with selection

Target Learners¶

• Researchers designing observational studies in education

Target Learning Objectives¶

• Reducing selection bias when estimating effects of educational interventions on achievement

Claims¶

Related Theories¶

Examples¶

• Three-dataset within-study comparison design varying pretest-selection correlation

Key Sources¶

• Kelly Hallberg, Thomas D. Cook, Peter M. Steiner, M. H. Clark. (2018). Pretest Measures of the Study Outcome and the Elimination of Selection Bias: Evidence from Three Within Study Comparisons. Prevention Science, vol. 19, issue 3. https://www.mathematica.org/publications/pretest-measures-of-the-study-outcome-and-the-elimination-of-selection-bias-evidence-from-three