SteppedPower 0.4.0

• glsPower() now supports count outcomes via family="poisson"
• Added ICC (intracluster correlation) transformation functions: icc_to_RandEff(), RandEff_to_icc(), RandEff_to_alpha012(), and alpha012_to_RandEff() for converting between random effects variances and ICC/CAC/IAC parameters
• Added vignette on binomial and count outcomes, with pre-calculated contour plots
• In glsPower(), the argument N now overrides the N stored in a supplied DesMat object
• glsPower() now fails gracefully (with a warning) if the information content cannot be calculated
• Character input arguments (e.g. dsntype, family) now throw an error if no known option is sufficiently similar, instead of silently choosing the closest match
• Diagnostic output now uses message() instead of print()
• Vignette plots now use the plotly partial bundle to reduce package size
• Changed covariance matrix construction to use fbdiag (fast block diagonal matrix)
• plot_CellWeights() now treats NA entries in incompMat as unobserved cluster periods
• Replaced \() with function() for backward compatibility with older R versions
• Fixed typo in RandEff_to_alpha
• Added tests for construct_DesMat(), construct_CovMat(), and glsPower()
• Updated vignettes and improved documentation with additional links in help files
• Roxygen documentation now uses markdown format; re-roxygenised all documentation

SteppedPower 0.3.5

• Addressed CRAN comments
• Fixed roxygen package name bug

SteppedPower 0.3.4

• Addressed CRAN comments

SteppedPower 0.3.3

• N (subjects per cluster-period cell) now belongs to DesMat class
• Added vignette for incomplete designs
• Fixed bug for handling NA in incompMat and trtMat

SteppedPower 0.3.2

• The most noticeable change in this version is that the abbrevation wls (weighted least squares) in function names is now replaced with gls (generalised least squars) to more properly reflect the scope of the functionality. For example, the function wlsPower() is now called glsPower() - although the former version still works and throws a warning.
• The closed formula for the computation of information content is now a dedicated formula, called compute_InfoContent()
• In plot.glsPower() there now is an option to manually set the font size of the annotation in the influence plots

SteppedPower 0.3.1

• The function wlsPower() now also computes the information content of cluster-period cells. Computation is currently done twice, once with a general formula and once explicitly. Information content of whole periods or clusters is also computed.
• The method plot.wlsPower() recieved multiple updates:
  • It now produces up to four plots: the projection matrix, the information content, the intervention design and the covariance matrix.
  • Incomplete designs (SWD where some cluster-period cells are omitted) are now visualised
  • Plots of projection matrix and information content can now be annotated with particular values in each cell; This is the default for smaller designs and can be turned on/off via annotations = <TRUE/FALSE>
  • An option show_colorbar to hide colour bars was added
  • An option marginal_plots to hide marginal plots on whole periods or clusters was added.
  • Various aesthetic improvements, e.g.: Improved hover information, dynamic gap size between cells.
• Vignette extended

SteppedPower 0.2.0

• The function wlsPower() now has an argument alpha_012 that offers an alternative way to specifiy the correlation matrix.
• In function wlsPower(), the argument AR now accepts a vector of up to three values. This allows to specifiy autoregressive structures for only a subset of: random cluster intercept, random intervention effect and random subject intercept.
• Closed formulae were added.
• The method plot.wlsPower now produces up to three plots, the projection matrix, the intervention design and the covariance matrix.
• The vignette was extended.

SteppedPower 0.1.0

• Initial submission to CRAN