randomized control trials
Experiments that randomly assign subjects to treatment and control groups to isolate causal effects; increasingly used in development economics.
As defined in Patrick Collison
Where it comes up · 8
- James Heckman on Facts, Evidence, and the State of Econometrics2 claims
- Dani Rodrik1 claim
- Daron Acemoglu1 claim
- Patrick Collison1 claim
- Lant Pritchett on Education in Poor Countries1 claim
- Omer Moav on the Emergence of the State1 claim
- 466 - What Is Technology Doing To Us1 claim
- 352 - Hubris Chaos1 claim
Most important claims · 9 of 9Search all claims for “randomized control trials” →
- Unconditional cash transfers via mobile money are the most powerful multi-dimensional poverty intervention: giving roughly $700 directly to a household, or about $70,000 to a village, achieves what traditional NGOs spent millions trying to do (fixed roofs, toilets, doubled livestock, kids in school) because it puts power in villagers' hands to meet their own individually-varying needs, and the most efficient programs deliver 92 cents of every dollar directly because they avoid the staff, offices, and vehicles of traditional charities.352 - Hubris Chaos
- Economics over the last 20-30 years has been dominated by an 'identification revolution' prizing the identification of causal effects (e.g., randomized controlled trials), but identifying a causal effect is a means, not the goal; the goal is to understand reality, and the profession has to a large extent lost interest in logical claims and theory.Omer Moav on the Emergence of the State
- Every research tool is useful but should not be overused: randomized controlled trials are not useless, but expecting them to answer the majority of development-economics questions is even more wrong; similarly cross-country growth regressions were hugely overrated yet remain indispensable for the overwhelmingly important questions of national development.Daron Acemoglu
- Much of the assault on universities has been directed at uncontroversial sciences—biology, medicine, physics—not just politicized humanities; defunding work like radar-invisible fighter-jet coatings, halting midstream cancer trials, stopping quantum and cryptography research, and freezing international talent constitutes irrational self-immolation of American science.466 - What Is Technology Doing To Us
- The fashionable use of randomized control trials (RCTs) in development economics, while possibly pointing in the right direction, risks feeding the spider system's 'input fantasy' that discovering the right combination of inputs will solve the problem, because it ignores the deeper system, accountability, and motivation problems and there is no evidence to date that experimentally identified interventions will replicate at anything like the needed scale of impact.Lant Pritchett on Education in Poor Countries
- Randomized control trials in social settings are systematically compromised because people denied access to a treatment act purposefully and seek substitutes—Head Start controls enrolled in other childcare programs, denied job-trainees took other programs, AIDS-trial subjects shared medicine with controls—so a simple treatment-versus-control comparison understates the true impact of the intervention.James Heckman on Facts, Evidence, and the State of Econometrics
- A microlending intervention in part of India showed no effect because later analysts discovered roughly 40 comparable lending programs already in place, making them near-perfect substitutes for the treatment and completely compromising the randomized trial.James Heckman on Facts, Evidence, and the State of Econometrics
- Randomized control trials were a fantastic contribution because they identify causal effects well (strong internal validity), but they are poor at explaining why effects occur or whether they generalize to other contexts (weak external validity); they now crowd out conceptual and theoretical work, overcorrecting from an earlier era when empirical development economics could not get you a job.Dani Rodrik
- Randomized control trials, though expensive and limited to narrower questions, are remarkable because they produce actual knowledge that is probably true; it is odd to turn down that reward even if you cannot use them to answer big-picture political economy questions that no one can steer.Patrick Collison