2017
DOI: 10.22237/jmasm/1493596980
|View full text |Cite
|
Sign up to set email alerts
|

Experiment-wise Type I error rates in nested (hierarchical) study designs

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1

Citation Types

0
3
0

Year Published

2020
2020
2023
2023

Publication Types

Select...
3

Relationship

0
3

Authors

Journals

citations
Cited by 3 publications
(3 citation statements)
references
References 13 publications
0
3
0
Order By: Relevance
“…The Benjamini-Hochberg (B-H) procedure was used to minimize the likelihood of false effects due to multiple tests (Benjamini & Hochberg, 1995). The procedure was applied to follow-up tests when the analysis was not a hierarchical model, as hierarchical analyses are sufficiently conservative (Gelman & Hill, 2007) and such corrections significantly limit power (Sawilowsky & Markman, 2017).…”
Section: Resultsmentioning
confidence: 99%
“…The Benjamini-Hochberg (B-H) procedure was used to minimize the likelihood of false effects due to multiple tests (Benjamini & Hochberg, 1995). The procedure was applied to follow-up tests when the analysis was not a hierarchical model, as hierarchical analyses are sufficiently conservative (Gelman & Hill, 2007) and such corrections significantly limit power (Sawilowsky & Markman, 2017).…”
Section: Resultsmentioning
confidence: 99%
“…The H ‐test was used to test whether the distribution of stem density and basal area differed between stands. For statistically significant results ( p < 0.05), Dunn's Kruskal–Wallis multiple comparisons were used with Holm's p ‐adjusted method and a correction to control the experiment‐wise Type I error rate (trueε̂ 2 ) to accurately determine difference between stands (Sawilowsky & Markman, 2017). We used an analysis of covariance (ANCOVA) model to adjust for the effect of varying time and site covariates on post‐sunset temperature (Oakes & Feldman, 2001).…”
Section: Methodsmentioning
confidence: 99%
“…For all our inferential tests, we used an alpha threshold of .05 without correction for family-wise errors in light of our (conditional) hierarchical approach (Sawilowsky & Markman, 2017) and the absence of union-intersecting testing (Parker & Weir, 2020). We opted for a data-driven, stepwise approach to our profiling analyses because we could not rely on earlier multilayer-multivariate research in this area to guide our variable selection and because we met key conditions for stepwise to lead to correct model choices (i.e., large but not too large number of predictors, no multi-collinearity, large sample size; Ganeshanandam & Krzanowski, 1989;Olejnik, Mills, & Keselman, 2000).…”
Section: Multivariate Profilingmentioning
confidence: 99%