Going beyond simple sample size calculations: a practitioner's guide

McConnell, Brendon; Vera‐Hernández, Marcos

doi:10.1920/wp.ifs.2015.1517

Cited by 23 publications

(19 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The intra‐class correlation with our outcomes is equal to 0.05 for the sample of boys and 0.10 for the sample of girls. With a power of 0.8, the minimum detectable effect is 0.5 of a standard deviation (McConnell and Vera‐Hernandez, ; Spybrook et al ., ). Nonetheless, we obtain statistically significant results with the limited power of 0.4.…”

Section: Estimation Strategy and Resultsmentioning

confidence: 99%

The curse of low expectations

Battaglia

Lebedinski

2017

Econ Transit

View full text Add to dashboard Cite

We examine how a remedial education programme for primary school-age children affects parental expectations about their children's future. Using original survey data we collected in Serbia, we investigate whether expectations on labour market prospects and educational attainment change as a consequence of exposure to the Roma Teaching Assistant programme. Our results show that parents of pupils in treated schools expect higher returns to education for their children and are more likely to expect them to achieve a secondary level of education. We also investigate the possible mechanisms in place due to the characteristics of the programme: remedial education and role model.

show abstract

Section: Estimation Strategy and Resultsmentioning

confidence: 99%

The curse of low expectations

Battaglia

Lebedinski

2017

Econ Transit

View full text Add to dashboard Cite

show abstract

“…The minimal sample size n 0 in N could, for example, be determined by power calculations (see, e.g. Duflo, Glennerster, and Kremer, 2007;McConnell and Vera-Hernández, 2015) that guarantee a certain power level for an hypothesis test of β = 0.…”

Section: Discussionmentioning

confidence: 99%

“…There is a large and important body of literature on the design of experiments, starting with Fisher (1935). There also exists an extensive body of literature on sample size (power) calculations; see, for example, McConnell and Vera-Hernández (2015) for a practical guide. Both bodies of this literature are concerned with the precision of treatment effect estimates, but neither addresses the problem that concerns us.…”

Section: Introductionmentioning

confidence: 99%

Optimal data collection for randomized control trials

Carneiro

Lee

Wilhelm

2019

The Econometrics Journal

View full text Add to dashboard Cite

In a randomized control trial, the precision of an average treatment effect estimator and the power of the corresponding t-test can be improved either by collecting data on additional individuals, or by collecting additional covariates that predict the outcome variable. We propose the use of pre-experimental data such as a census, or a household survey, to inform the choice of both the sample size and the covariates to be collected. Our procedure seeks to minimize the resulting average treatment effect estimator's mean squared error or the corresponding t-test's power, subject to the researcher's budget constraint. We rely on a modification of an orthogonal greedy algorithm that is conceptually simple and easy to implement in the presence of a large number of potential covariates, and does not require any tuning parameters. In two empirical applications, we show that our procedure can lead to substantial gains of up to 58%, measured either in terms of reductions in data collection costs or in terms of improvements in the precision of the treatment effect estimator.JEL codes: C55, C81.

show abstract

Section: Discussionmentioning

confidence: 99%

Optimal data collection for randomized control trials

Carneiro

Lee

Wilhelm

2017

View full text Add to dashboard Cite

In a randomized control trial, the precision of an average treatment effect estimator and the power of the corresponding t-test can be improved either by collecting data on additional individuals, or by collecting additional covariates that predict the outcome variable. We propose the use of pre-experimental data such as other similar studies, a census, or a household survey, to inform the choice of both the sample size and the covariates to be collected. Our procedure seeks to minimize the resulting average treatment effect estimator's mean squared error or the corresponding t-test's power, subject to the researcher's budget constraint. We rely on a modification of an orthogonal greedy algorithm that is conceptually simple and easy to implement in the presence of a large number of potential covariates, and does not require any tuning parameters. In two empirical applications, we show that our procedure can lead to reductions of up to 58% in the costs of data collection, or improvements of the same magnitude in the precision of the treatment effect estimator.

show abstract

Going beyond simple sample size calculations: a practitioner's guide

Cited by 23 publications

References 37 publications

The curse of low expectations

The curse of low expectations

Optimal data collection for randomized control trials

Optimal data collection for randomized control trials

Contact Info

Product

Resources

About