NEWS

Durga 2.0.0.9000

DurgaDiff takes a parameter ef.size.cex that controls the size of the effect size central tendency symbol.

DurgaBrackets prevents bracket text from overlapping.
DurgaDiff now calculates CI of group mean for groups with at least 3 values and 2 distinct values. Previously it was only calculated for at least 3 distinct values.
DurgaDiff now correctly handles R = NA as documented; does not calculate CIs

DurgaDiff can now calculate multiple different effect sizes. The old values for effect.type of "cohens" and "hedges" are no longer accepted and generate an error. This backwards-incompatible change has been made so that users are required to consider the effect type they wish to use. See the help file for DurgaDiff for more information.
DurgaDiff now calculates CI of group means using BCa bootstrapping rather than using a parametric formula. This means that a group with fewer than 3 distinct values will not have a CI calculated (lower and upper limits will be NA).
DurgaDiff now accepts a vector for group.col. A single group column is created by concatenating values from each specified group column.
DurgaDiff now accepts data in either long or wide format. Added the data set insulin.wide as an example of wide format.

DurgaPlot: Ensure entirety of effect size is visible when on the right.
DurgaDiff: bootstrap implementation of CI samples each group separately. The bug sometimes meant that the effect size axis disappeared in plots for groups with small sample sizes. The confidence intervals will be slightly different now.

Improvements to vignette and online help.
Added argument violin.params to DurgaPlot to provide additional control over violin appearance.
Added arguments ef.size.lty and ef.size.lwd to DurgaPlot to provide control over effect size error bar appearance.
Added argument points.adjust to DurgaPlot to allow control over point layout.
Bug fix in DurgaPlot: violin.fill = FALSE now correctly does not fill violins, rather than white fill.
Bug fix in DurgaBrackets; drawing many brackets sometimes resulted in overlapping brackets.
Extend tick marks on right axis to cover CI when plotting standardised effect size on the right.
Fixed bug in plotting violins for pathological data; previously failed with error "Error in xy.coords(x, y, setLab = FALSE) : 'x' and 'y' lengths differ"