English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

This paper studies inference in randomized controlled trials with multipletreatments, where treatment status is determined according to a "matchedtuples" design. Here, by a matched tuples design, we mean an experimentaldesign where units are sampled i.i.d. from the population of interest, groupedinto "homogeneous" blocks with cardinality equal to the number of treatments,and finally, within each block, each treatment is assigned exactly onceuniformly at random. We first study estimation and inference for matched tuplesdesigns in the general setting where the parameter of interest is a vector oflinear contrasts over the collection of average potential outcomes for eachtreatment. Parameters of this form include standard average treatment effectsused to compare one treatment relative to another, but also include parameterswhich may be of interest in the analysis of factorial designs. We firstestablish conditions under which a sample analogue estimator is asymptoticallynormal and construct a consistent estimator of its corresponding asymptoticvariance. Combining these results establishes the asymptotic exactness of testsbased on these estimators. In contrast, we show that, for two common testingprocedures based on t-tests constructed from linear regressions, one test isgenerally conservative while the other generally invalid. We go on to apply ourresults to study the asymptotic properties of what we call "fully-blocked" 2^Kfactorial designs, which are simply matched tuples designs applied to a fullfactorial experiment. Leveraging our previous results, we establish that ourestimator achieves a lower asymptotic variance under the fully-blocked designthan that under any stratified factorial design which stratifies theexperimental sample into a finite number of "large" strata. A simulation studyand empirical application illustrate the practical relevance of our results.

Inference for Matched Tuples and Fully Blocked Factorial Designs