NEWS

semPower 2.2.0

single group/occasion identification is now the default in models assuming metric or scalar invariance in powerMI and powerLI. The previous behavior can be obtained by setting singleGroupIdent = FALSE / singleOccasionIdent = FALSE.
allow referent scaling in powerMI and powerLI
allow multifactor models at each measurement in powerLI
allow definition of effect size in powerRegression in terms of R squared
more sensible structure of returned simulated data in the multigroup case
add option to specify a specific strategy to resolve futures when using parallel processing to estimate simulated power
bugfix: make simulated power with robust test-statistics (also) work with lavaan > 0.20.
bugfix: proper evaluation of parameter bias in simulated power when models involve a single factor or single regression relations.
bugfix: respect requested autocorrelated residuals without invariance constraints in powerARMA, powerAutoreg

update to R 4.5.0
add option to estimate direct effects with population values of zero in powerMediation
bugfix: fix genSigma with supplied means when Lambda includes secondary loadings

bugfix: make simulated power work properly for multigroup models with > 2 groups

update to R 4.4.0
bugfix: don't rely on dummy latents in powerMediation with only observed variables. This led to underestimated effects and thus underestimated power.
bugfix: fix edge cases with variable ordering in user-provided Sigmas
bugfix: make definition of factor model via Lambda in powerBifactor work