| NEWS | R Documentation |
properly adjusted imports of functions/methods from other packages
dropped dependencies on polycor and psychometric, item analysis is now
bone using psych::alpha()
removed outdated internal klausuR-internal.roxy.all.R file
the option 'keep.cases' in klausur.data() was broken
klausur.report(): fixed LaTeX issue if percent sign was used in psedonyms
klausur.report(): global reports are now printed in landscape mode
reaktanz.de supports HTTPS now, updated references
the source code has moved from R-forge to https://github.com/unDocUMeantIt/klausuR
new argument "fancyhdr" in klausur.report() to add information to headers
and footers of generated reports
cleaned up klausur.report(): moved internal functions to separate file
adjusted the auto table sizes in klausur.report() a bit
rephrased the description to comply with CRAN terms
support for pick-n scoring was added to klausur()
new slot "distractor.analysis" in objects of class klausuR, listing the
selected answer alternatives for each individual item and some statistics
like correlation with the overall results and mean results of subjects
chosing an alternative
new element "wrong" in slot "corr" of klausur.answ class items for pick-n
scoring
new argument "wrong" in klausur.data() for the same reason
klausur.report() gained options from sort() for "anon" and "glob" results
new elements "keep.cases" and "recode.na" in klausur.data(), to keep
cases with missing data, even if "na.rm=TRUE"
stripping trailing whitespace from answer data if it is in factor of
character format. this happens easily if data was imported from SPSS and can
cause nasty problems, because the whitespace characters are interpreted by
klausur() as a valid but wrong answer
removed typos in klausur.repost() code ("appen" instead of "append")
anonymous feedback output is now sorted by points as a default. the previous default took the order as is, which usually sorted it by family names of participants, which limits anonymity to a certain degree
the character "'" is now also filtered out of file names
breaks for axis labels in histograms are now calculated by internal
function axis.breaks(), which fixes a problem with non-integer points
klausur.report() does now also check for digits to display in global and
anonymous feedback PDF
klausur.report() now shows empty answers ("") as empty, not as a question
mark
LaTeX replacements of characters has been corrected to also match first character
manual setting of "table.size" was buggy in klausur.report(), adding just
the word of the chosen size to the table
klausur.mufo() was using outdated object classes
polycor was added as a dependency (needed for distractor analysis)
some code cleaning in klausur() for better readability
added a proper GPL notice at the beginning of each R file
replaced all tabs in the source code by two space characters
moved the internal helper function find.partial() out of the function
body of partial() for cleaner code
show method for klausuR results now also reports distractor analysis, if found
partial points are no longer rounded before end results are calculated
pdfLaTeX became picky and didn't work with the table of marks any more.
set print.xtable() to use floating=FALSE to fix this.
added some more character filtering on generated filenames which could cause LaTeX to fail with an error
no real new features, restructured the source code layout, as roxygen2 doesn't seem to get collate info easy with R 3.0.0 ...
corrected LaTeX output (carets)
added missing date stamps in ChangeLog for releases prior to version 0.12-1, using the repo archive
item analysis now includes Lienert's selection index (selektionskennwert)
debianized sources
check for missing items in data and/or correct answers
force item names into expected letter case
corrected LaTeX output (filtering of problematic characters)
replaced custom function to clean up LaTeX code with
encoded_text_to_latex() from tools package in klausur.report()
replaced UTF-8 characters with \uxxxx escapes in klausur.report()
set some internal variables NULL in compare(), grand.table(),
klausur.gen(), klausur.mufo(), klausur.report() and the internal function
klausur.reorderItems(), to avoid needless NOTEs from "R CMD check"
re-formatted the ChangeLog to have the roxyPackage translate it into a NEWS.Rd file
fixed bug in compare(): "new.set" didn't work if exactly one or none
differences were found
added 'maxp' option to klausur.answ-class and klausur.date(), to be able
to compensate for really buggy tests...
adjusted klausur() to check for maxp
added option "merge" to klausur.report(), to merge all individual reports
into one file
added status feedback to klausur.report(), can be turned of with new
option "quiet=TRUE"
updated the RKWard GUI to the new options
fixed bug in nret.rescale() regarding NR data
fixed author definition for R 2.14
cleaned the docs so package check runs fine
fixed typo in klausur.report()
0.12 is not fully downward compatible with earlier versions of the klausuR package!
drastic changes: klausur() and klausur.mufo() from now on only take
objects of the new class klausur.answ, which replaces the 'answ' and 'corr'
arguments completely. Most of the time it took to evaluate tests was
squeezing data with exotic variable names and loads of surprises into a valid
data.frame. Those days should finally be numbered now...
added function klausur.data() to do the squeezing into klausur.answ class
objects (s.o.)
added scoring function NRET+, which is more conservative than NRET when it comes to errors
added function nret.rescale() to change scaling of points in result
objects (NR/ET/NRET/NRET+)
added function grand.table() to export all results back to CSV format
the class klausuR gained a new slot 'misc' to preserve whatever unused variables are left over from the squeezing process.
introduced class klausuR.test, now as a placeholder for future releases, and added slots to classes klausuR and klausuR.answ
added function nret.translator() to convert NRET/ET data from/to SPSS
added options 'table.size' and 'NRET.legend' to klausur.report()
smarter handling of histograms and marks info in klausur.report()
klausur.report() can now also create histograms with data from several
klausuR objects
klausur.report() can now produce reports from more than one object, if
you give them in a list
klausur.compare() was renamed to compare() and can now cope with NAs,
unequal cases and unequal variables (will cause verbose warnings), and rename
input variables, if necessary. it is also no longer limited to compare
klausuR data, hence the renaming.
fixed some NRET/ET related bugs in the plot method and klausur.gen.marks()
fixed 'if' statement error in show method for klausuR-class objects
added support for item analysis on ET/NRET data
added NRET sample data "spss.data"
added SD values to item analysis (show method)
added 'sort.by' option to klausur.data(), and klausur()
added sort method for objects of class klausuR
added the full RKWard plugin as inst/rkward, so both get updated simultanously
adapted rkwardtests to current behaviour of klausuR
sources can now be roxygenized by roxygen2 (classes docs are static until '@slot' is supported again)
aligned penalty for false negatives to number of correct alternatives in NRET/ET scoring
for NRET/ET, moved determination of max. points etc. to new internal
function nret.minmax()
added internal function roxy.package() to ease development
added initial support for elimination testing (ET) and number right elimination testing (NRET)
fixed "Form" addition in klausur.gen()
fixed bug in points (were logical instead of numeric, which caused trouble for PDF generation)
fixed annoying "show" warning
complete rewrite of klausur.compare(): less and more efficient code
replaced "na.replace" option with a plain "na.rm", i.e., NA cases are
dropped completely by default
added new "na.rm" option to plot method, TRUE by default
added variable "Form" to klausur.gen()
replaced several "if(a == b)" constructs with "identical(a, b)" or "isTRUE(...)"
fixed: klausur.report() accidently created PDFs in a loop over and over
again
fixed: klausur.report() failed if "#" appeared in names
fixed: klausur.report() didn't show partial results correctly, added
check for needed digits
added slots "wght" and "points" to class klausuR
added klausur.mufo() and new S4 class klausuR.mult: initial support for
parallel test forms
added show() and plot() methods for objects of class klausuR.mult
added support for objects of class klausuR.mult in klausur.report()
fixed missing import of tools package
dropped roxygen dependency (still using roxygen, though)
some code cleanup for better readability
added direct PDF creation to klausur.report(), using texi2dvi() from
package tools
consequently added the tools package as a dependency
added schemes "DIHK", "USA" and "UK" to klausuR.gen.marks()
added informative slot "marks.sum" to objects of class klausuR
updated the show method for klausuR objects accordingly
added "marks.info" option to klausur.report(), to include all or parts of
"marks.sum" if wanted
fixed another bug in klausur.compare(): objects with NAs meant trouble.
added "select" option to klausur.compare() to be able to focus on certain
variables
added support for different file name scheme to klausur.report() (matr.
no. or name_firstname)
added "alt.comma" parameter to klausur.report()
minor corrections in the documentation (some missing params in usage)
restructured the LaTeX output
fixed some bugs in klausur.compare()
added parameters to skip cronbach's alpha and item analysis in klausur()
cronbach's alpha and item analysis now run in a safe tryCatch()
environment
modified the show() method accordingly
added "alpha if removed" calculation for each item
added longtable support for klausur.report()
fixed some bugs in klausur(), klausur.report() and plot() regarding
character vectors
added support for partially correct answers (if multiple correct answers
possible) via the new "score" parameter
more cleaning up: moved some subroutines to internal functions
added marks A to F as an option in klausur.gen.marks()
adjusted the plot method to treat marks as characters
beautyfied the output of individual results in klausur()
cleaned up a bit (removed unused relicts from klausur.R)
translated hopefully all comments in the code into english
added feature to suggest marks by normal distribution
initial public release (via reaktanz.de)