Changes in version 0.4.2 (2025-08-18) Bug fixes - Fixed Rd cross-reference NOTE by adding package anchors to \link{} and \linkS4class{} targets. - Updated documentation for runEquateObserved() and SingleGroupClass to ensure proper linking to external packages. Other - Minor formatting updates to .Rd files. Changes in version 0.4.1 (2023-02-03) New features - Added support for working with GPC models. - Added support for when items between instruments have different item models. (e.g., the anchoring instrument uses GR, and the instrument being calibrated uses GPC.) - Added support for when items within the anchoring instrument use different item models. (e.g., the anchoring instrument uses GR for some items and GPC for others.) - Added support for when items within the instrument being calibrated use different item models. (e.g., the instrument being calibrated uses GR for some items and GPC for others.) Bug fixes - Fixed where loadData() would not warn about items that may need reverse coding for scale IDs other than 1. - Now requires TestDesign (>= 1.5.1). - Response probabilities are now computed faster using cpp functions. - Deprecated guiPROsetta() is removed. Use PROsetta() instead. - Updated documentation. - Made minor updates to the Shiny app in PROsetta(). - Removed ncat column in anchor parameters. This is now inferred from the number of parameters. - Added cpp routine for EAP computation after Lord-Wingersky recursion. This improves the speed of runRSSS(). - getRSSS() now nudges the user if the prior mean input looks like a T-score, which should be entered in the theta metric. - Updated CITATION to use bibentry() to meet CRAN requirements. Changes in version 0.3.5 (2022-01-03) - runRSSS() output now includes linear approximation betas when the CPLA method is used. - runLinking() output now includes the latent mean and variance when the FIXEDPAR method is used. - runCalibration(), runLinking(), runEquateObserved(), getCompleteData() gains verbose argument for printing status messages. Status messages that were used to be printed in previous versions are now suppressed by default. Changes in version 0.3.4 (2021-11-23) - Removed unused columns (min_score, reverse, scores) in example datasets for clarity. The package functions do not use these columns. - loadData() now warns if there is a variable that may need reverse coding. This is triggered by a negative correlation value. Changes in version 0.3.2 (2021-07-20) - Fixed where runLinking(method = "FIXEDPAR") was not working when the anchor instrument ID was not 1 in item map. - Fixed where runLinking(method = "FIXEDPAR") was not working when the anchor and target instruments had different numbers of categories in response data. - Fixed where runFrequency() was not sorting categories correctly when the number of categories was 10 or above. - Fixed where runCalibration(fixedpar = TRUE) was not reading anchor parameters correctly when an integer value existed in anchor parameters. - Fixed where item parameters for dichotomous items were triggering an error while being parsed. - For compatibility with R < 4.0, loadData() now sanitizes input data when a data frame is supplied. Changes in version 0.3.0 (2021-04-13) New features - runLinking() now supports method = 'CPFIXEDDIM' to perform two-dimensional calibration, for use in performing calibrated projection (Thissen et al., 2015). The difference with method = 'CP' is that 'CPFIXEDDIM' constrains the mean and the variance of the latent anchor dimension, instead of constraining anchor item parameters. For this purpose, a unidimensional fixed parameter calibration using only the anchor response data is performed to obtain the mean and the variance. - getRSSS() for computing a single raw-score to standard-score table is now exposed. Changes in version 0.2.1 (2020-11-25) QoL updates - Added getResponse() for extracting scale-wise response data from a PROsetta_data object. - Added getItemNames() for extracting scale-wise item names from a PROsetta_data object. Changes in version 0.2.0 (2020-09-20) New features - runLinking() now supports method = 'CP' to perform two-dimensional calibration, for use in performing calibrated projection (Thissen et al., 2011). - runLinking() now supports method = 'CPLA' to perform two-dimensional calibration, for use in performing linear approximation of calibrated projection (Thissen et al., 2015). - runRSSS() now performs two-dimensional Lord-Wingersky recursion with numerical integration, when the output from runLinking(method = 'CP') is supplied. - runRSSS() now performs linear approximation of calibrated projection, when the output from runLinking(method = 'CPLA') is supplied. - Shiny application PROsetta() now supports calibrated projection and its linear approximation. Bug fixes - runEquateObserved(type_to = "theta") now works. - loadData() now checks for a valid @scale_id. Changes in version 0.1.4 (2020-07-01) - First public release. Structural changes - PROsetta_config class and createConfig() are now deprecated. The functionalities are merged to PROsetta_data class and loadData(). - run*() functions now require PROsetta_data objects instead of PROsetta_config objects. - runLinking() now has method argument to specify the type of linking to perform. Accepts MM, MS, HB, SL, and FIXEDPAR. Updates - runLinking() is now capable of performing fixed calibration. - runCalibration() now performs free calibration by default. - runCalibration() and runLinking() now errors when iteration limit is reached, without returning results. - runRSSS() now returns thetas in addition to T-scores, and also expected scores in each scale. - runEquateObserved() now has type_to argument to specify direct raw -> T-score equating or regular raw -> raw equating. - Functions are now more verbose. - Added PROMIS Depression - CES-D linking dataset data_dep. - Added plot() for drawing raw score distribution histograms. - Added plotInfo() for drawing scale information plots. - Added several helper functions. - Made cosmetic improvements on Shiny app. Bug fixes - Fixed where Shiny app was displaying SL method linear transformation constants regardless of specified linking method. Changes in version 0.0.4 - Added scalewise argument to runClassical() and runCFA(). When TRUE, analysis is performed for each scale. - runEquateObserved() now removes missing values to produce correct raw sums. - loadData() now retains missing values. Changes in version 0.0.3 - loadData() now removes missing values.