NEWS (original) (raw)

eatRep 0.15.2

bugfix in the presentation of the regression tablerepGlm()
bugfix in sample size description on repTable() forengine = BIFIEsurvey
The direction of the group comparisons has been standardized so that the difference is reference group - focus group

bugfix in the presentation of the regression results (reporting function for repGlm())
bugfix in the naming of the label column in the plain sheet of the reporting function report2()

multicore computation is supported for all analyses using the survey package
computation of group differences using ‘group.differences.by’ now also works for adjusted means
for weighted analyses, the eatRep functions now remove zero weight cases prior to analyses to avoid counting them when determining sample size
the eatRep functions now issue more informative alerts if missings occur on dependent/independent or jackknife variables. Missing occurrences concerned as crucial will be caught with an error message
enhanced warning messages if number of nests/imputations differ between grouping variables
enhanced warning messages if number of jkrep units differ between nests/imputations
bugfix in the cross-level differences of group differences computation
bugfix if levels of grouping variables contain leading and/or trailing spaces
new function report2 which summarizes the output of the four main functions a little tidier and provides an interface for eatPlot. The old reporting function report() is deprecated
new function pool.R2 computes pooled R^2 for multiple imputed and nested multiple imputed regression analyses according to Harel (2009)

exemplary with variable labels stored as attributed
add ‘jkfac’ argument to functions which call BIFIEsurvey functions
trend analyses (with warning messages) now are possible when levels of grouping variables differ between assessment cycles
bug fix in the internally used function which checks linking error object for consistency
bug fix in repLmer when no weights are defined
reporting function was slightly simplified (i.e., remove additional target formats)

add function repLmer() for replication methods for linear multilevel models (wrapper toBIFIEsurvey::BIFIE.twolevelreg())
bug fix in internally used check function whether levels of grouping variable(s) are equal across assessment cycles (trend groups)
bug fix in singularity treatment of repGlm()
bug fix in repQuantile()
add Ns (sample size) to the output of repTable()function
add tests for repQuantile() andrepLmer()

sorting of parameters and determination coefficient enhanced in reporting function for repGlm()
add function checkLEs which checks consistency of linking errors with data which stem from an eatGADS data base
trend estimation was broadened for more than two measurement occasions
add class identifier in exemplary data

add CR0 and CR2 methods for weighted effect coding with heterogeneous variances in repMean() andrepGlm(), including a cluster argument
add class ID variable to example data set
bug fix: fixed inadequate column definition in the reporting function for repQuantile()
adapt repQuantile function for rewritten function svyquantile from the survey package
slightly revised vignette

add asterisks for significance in the console output of the reporting function for repGlm()
bug fix: fixed inadvertent warning message in function for group checks
bug fix: call svyby with return.replicates = FALSE for quantile function

fixed report for cross differences when levels of grouping variable contained the variable name
fixed missing entries in mode column in the output ofrepMean()
eatRep::repMean(...) now works as well aslibrary(eatRep); repMean(...)
added tests for repGlm() function