Random Control Trials

AKA: RCT

Evaluating the replicability of social science experiments in Nature and Science between 2010 and 2015

CHANNELLING FISHER: RANDOMIZATION TESTS AND THE STATISTICAL INSIGNIFICANCE OF SEEMINGLY SIGNIFICANT EXPERIMENTAL RESULTS

Warnings

  • Dicing RCT results up by coverates or in with a regression model instead of doing a simple T test can generate spurious results from a few high leverage outlier observations (Young 2019).

References

Young, Alwyn. 2019. “Channeling Fisher: Randomization Tests and the Statistical Insignificance of Seemingly Significant Experimental Results*.” The Quarterly Journal of Economics 134 (2): 557–98. https://doi.org/10.1093/qje/qjy029.