Meta-Analysis of Studies with Non-Statistically Significant Unreported Effects

Novel method to unbiasedly include studies with Non-statistically Significant Unreported Effects (NSUEs) in a meta-analysis and . Briefly, the method first calculates the interval where the unreported effects (e.g. t-values) should be according to the threshold of statistical significance used in each study. Afterwards, maximum likelihood techniques are used to impute the expected effect size of each study with NSUEs, accounting for between-study heterogeneity and potential covariates. Multiple imputations of the NSUEs are then randomly created based on the expected value, variance and statistical significance bounds. Finally, a restricted-maximum likelihood random-effects meta-analysis is separately conducted for each set of imputations, and estimations from these meta-analyses are pooled. Please read the reference in 'metansue' for details of the procedure.


News

metansue v2.2 (Release date: 2018-07-11)

Changes:

  • Objects of class "nsue" include now the function to calculate the variance from the effect size, the function to multiply impute, and the function to back-transform. This allows users to create their own objects of class "nsue" for effect sizes not included in the package.
  • The funnel function now plots the studies with NSUEs as smooths dots representing several imputations.
  • For "visual" compatibility with standard meta-analyses, heterogeneity test now reports a Q statistic (instead of an F statistic).
  • Function "z_from_r" has been renamed as "zcor_from_r".
  • Adjustment for repeated measures and functions "r_in_smd_from_t_means_and_sds1" and "r_in_smd_from_t_means_and_sds2" have been discontinued, please use older releases of the package to use them.

metansue v2.1 (Release date: 2018-03-13)

Changes:

  • Improved calculation of the pooled covariance matrix.

metansue v2.0 (Release date: 2017-07-21)

Changes:

  • Several improvements in robustness. These changes imply that "data" argument is no longer required, that a single hypothesis can be specified in "meta", and that the following functions for "meta.nsue" objects have been removed: "linearHypothesis", "metalm" and "subset".
  • Added "subset" function for objects of class "nsue".
  • Default number of imputations has been increased to 500.

metansue v1.2 (Release date: 2016-08-31)

Changes:

  • Bugs fixed: study labels can be numeric, correlation forest can have no NSUEs.

metansue v1.1 (Release date: 2016-06-20)

Changes:

  • Added "r_in_smd_from_t_means_and_sds" and "subset" functions.

metansue v1.0 (Release date: 2015-11-11)

First version.

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.

install.packages("metansue")

2.3 by Joaquim Radua, 5 days ago


Browse source code at https://github.com/cran/metansue


Authors: Joaquim Radua


Documentation:   PDF Manual  


Task views: Meta-Analysis


GPL-3 license



See at CRAN