Analysis and Graphics for Unreplicated Experiments



unrepx package for R: Analysis of unreplicated experiments


Russ Lenth, University of Iowa, russell-lenth@uiowa.edu

The unrepx package provides much of what one needs to analyze basic unreplicated screening experiments such as two-level factorial and fractional factorial designs, in which one has a set of independent effect estimates but no degrees of freedom for error. The analyses are based on underlying assumptions that the effect estimates are independent normal, all with with the same variance; and that effect sparsity holds whereby only a few of the effects are active.

The package provides basic functions yates() and gyates() for estimating effects; plotting functions hnplot() (half-normal plot), refplot() (reference plot), and parplot() (Pareto plot) for displaying effects and depicting which may be active; and statistical analysis functions PSE() (pseudo standard error), ME() (margin of error), ref.dist() (reference distribution), and eff.test() (tests of effects). The dot.plot() function that underlies ref.plot() is also usable in its own right for creating resizeable dot plots. There are also example effect estimates, and a function for creating half-normal graph paper.

To see an overview of these features, use vignette("overview", package = "unrepx"). You may also view it here, but without the graphs


  • To install latest version from CRAN (once it is available there), run
  • To install the latest development version from Github, have the newest devtools package installed, then run
devtools::install_github("rvlenth/unrepx", dependencies = TRUE, build_vignettes = TRUE)

For latest release notes on this development version, see the NEWS file