Evaluation of the Power of Rerandomization Tests, With Application to Weather Modification Experiments

Gabriel, K. Ruben; Hsu, Chin-Fei

doi:10.2307/2288181

Cited by 6 publications

(3 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A re-randomization test (Gabriel, 1979(Gabriel, , 1981Tukey et al, 1978), which makes no assumptions about the distributions of the calculated statistics, was applied within each regime to test for the significance of the differences in these DSD metrics between the seeded and unseeded populations within the given aerosol regime. As noted by Gabriel and Hsu (1983), the test's lack of assumptions about the nature of the data set is an asset in weather experiments where the natural variability can be very high and largely uncontrolled, and the experimenters have to accept experimental units as they occur.…”

Section: Observational Analysis Methodsmentioning

confidence: 99%

The Influence of Hygroscopic Flare Seeding on Drop Size Distribution Over Southeast Queensland

et al. 2021

View full text Add to dashboard Cite

Hygroscopic seeding has been explored in various regions of the world with the goal of enhancing rainfall (Bruintjes, 1999; Flossmann et al., 2019; Silverman, 2003). The conceptual model of hygroscopic seeding focuses on increasing the precipitation efficiency of warm-based convective clouds by broadening the droplet spectra to enhance warm rain formation processes. The approach is to introduce large hygroscopic particles that act as efficient cloud condensation nuclei (CCN) or giant CCN (GCCN), which then nucleate large drops that grow more rapidly via condensation and initiate effective collision-coalescence processes. In addition, this process can suppress the activation of small background aerosols via the "competition effect" (Cooper et al., 1997; Ghan et al., 1998; O'Dowd et al., 1999). Even in mixed-phase, deep convective clouds, the concept suggests that more large drops and drizzle produced from warm rain processes would lead to more active mixed-phase ice formation and subsequent increases in the precipitation efficiency (Lawson et al., 2015; Rosenfeld & Woodley, 1993). Only a few studies have directly measured the broadening of the droplet spectra attributed to hygroscopic seeding (Ghate et al., 2007; Mather et al., 1997). While Rosenfeld et al. (2010) also attempted to directly measure droplet broadening from hygroscopic flare seeding, they were unable to detect a distinct change in the drop size distribution (DSD). However, their observational results were only based upon measurements from one seeded cloud. This study uses data collected in Southeast Queensland to examine the impact of hygroscopic seeding on the initial droplet spectra. A decade-long drought in Southeast Queensland, Australia, in the early 2000s, known as the Millennium Drought (van Dijk et al., 2013) prompted the Queensland government to seek ways to create more water resources, which led to a feasibility study to investigate precipitation enhancement via cloud seeding known as the Queensland Cloud Seeding Research Program (QCSRP; Tessendorf et al., 2010; 2012). Scientists from the United States, South Africa, and Australia conducted the feasibility study, which included a randomized hygroscopic cloud-seeding trial, during the summers in 2007-2009. Given the small number of cases and the high level of natural variability in the QCSRP randomized experiment sample, the results were incomplete and not statistically significant (Tessendorf et al., 2010). This is

show abstract

Section: Observational Analysis Methodsmentioning

confidence: 99%

The Influence of Hygroscopic Flare Seeding on Drop Size Distribution Over Southeast Queensland

et al. 2021

View full text Add to dashboard Cite

show abstract

“…This distinction is also followed by Cox and Hinkley (1974), Willmes (1987), Mewhort (2005), Zieffler, Harring, and Long (2011), and Keller (2012). Randomization tests (under the strict random assignment model) are also sometimes called "rerandomization tests" (Brillinger, Jones, & Tukey, 1978;Gabriel, 1979;Gabriel & Hall, 1983;Gabriel & Hsu, 1983;Petrondas & Gabriel, 1983), as an analogy with "resampling tests", but this terminology is rare in the current scientific literature. Furthermore, this terminology might be misleading because it contains a suggestion that a new test procedure is proposed (although it is an "ordinary" randomization test).…”

Section: Terminological Clarificationmentioning

confidence: 99%

Randomization Tests or Permutation Tests? A Historical and Terminological Clarification

Onghena¹

2017

Randomization, Masking, and Allocation Concealment

View full text Add to dashboard Cite

The terms "randomization test" and "permutation test" are sometimes used interchangeably. However, there are both historical and conceptual reasons for making a clear distinction between the two terms. Using a historical perspective, this chapter emphasizes the contributions made by Edwin Pitman and Bernard Welch to arrive at a coherent theory for randomization and permutation tests. From a conceptual perspective, randomization tests are based on random assignment and permutation tests are based on random sampling. The justification of the randomization test derives from the fact that under the null hypothesis of no treatment effect, the random assignment procedure produces a random shuffle of the responses. The justification of the permutation test derives from the fact that under the null hypothesis of identical distributions, all permutations of the responses are equally likely. It is argued that this terminological distinction is crucial for recognizing the assumptions behind each of the tests and for appreciating the validity of the corresponding inferences.

show abstract

“…For an extreme example, if N is chosen such that K < 20 by Equation (1), then Equation (15) implies that a 5% level randomization test will result in an error of the second kind with absolute certainty (probability of 100%) because in that case it is impossible to have for any true alternative hypothesis. Besides N, the statistical power of a randomization test (i.e., the complement of the probability of an error of the second kind) is function of the design, the vector of basic responses, the treatment effect, the test statistic, and the level of significance α (Gabriel & Hall, 1983;Gabriel & Hsu, 1983;Keller, 2012;Kempthorne & Doerfler, 1969). Because the most sensitive test statistic can be devised for the specific treatment effect that is expected, randomization tests with optimal power can be constructed in a variety of situations.…”

Section:  Validity and Power By Constructionmentioning

confidence: 99%

Randomization and the Randomization Test: Two Sides of the Same Coin

Onghena¹

2017

Randomization, Masking, and Allocation Concealment

View full text Add to dashboard Cite

This chapter provides a framework for conceptualizing randomization in clinical trials and for linking randomization to a statistical test. The aim of this chapter is to demonstrate how randomization works, how it does not work, and why a discussion of randomization without reference to a randomization test is incomplete. Randomization works because, on average, the differences between the treatment averages of any potentially confounding variable are zero. Randomization does not work in the sense that it does not contain any magic equating potion that spirits away all biases for any particular clinical trial. Furthermore, it should be taken into account that randomization is closely linked to statistical inference by the concept of a randomization test. This test is formally defined and its main features are highlighted: The randomization test is valid and powerful by construction, it can be used with any design and any test statistic, without random sampling and without assuming specific population distributions, and it results in a frequentist, conditional, and causal inference. The randomization test derives its statistical validity by virtue of the actual randomization and conversely a randomized design is complemented with a calculation of the randomization test p-value because it provides a quantification of the probability of the outcome (or a more extreme one) if the null hypothesis is true.

show abstract

Evaluation of the Power of Rerandomization Tests, With Application to Weather Modification Experiments

Cited by 6 publications

References 0 publications

The Influence of Hygroscopic Flare Seeding on Drop Size Distribution Over Southeast Queensland

The Influence of Hygroscopic Flare Seeding on Drop Size Distribution Over Southeast Queensland

Randomization Tests or Permutation Tests? A Historical and Terminological Clarification

Randomization and the Randomization Test: Two Sides of the Same Coin

Contact Info

Product

Resources

About