Changes in Version 0.6.1-0
o added the function pairwise.SepRel() to compute the Person Separation Reliability
Changes in Version 0.6.0-0
o added the data.frame with the Knox Cube Test Data from Wright & Stone (1979)
o added the option of MINCHI estimation for item parameters
Changes in Version 0.5.0-2
o fixed error message during re-building of vignette outputs
Changes in Version 0.5.0-1
o added S3 method 'residuals' for objects of class 'pers' resulting from function pers()
Changes in Version 0.5.0-0
o added estimation of thresholds for Rater severity / repeated measurement in function pair()
Changes in Version 0.4.5-1
o fixed the 'S4 exports' issue
Changes in Version 0.4.5-0
o added a summary method for objects of class 'q3' resulting from function q3() for Yens (1984) Q3 fit statistic
o fixed a bug in the calculation of the max and min values in the Q3 statistic (thanks to Moritz Heene! for the hint)
Changes in Version 0.4.4-6
o fixed the '`class(.) == *`' issue
o changed pairwise specific object class for results from functions pairwise.person.fit() and pairwise.item.fit() to 'pifit' and 'ppfit' respectively
o changed pairwise specific object class for results from function pairwise.S() 'pairS'
Changes in Version 0.4.4-5.3
o fixed a minor bug in (deprecated) internal function 'pvx.super'
Changes in Version 0.4.4-5.2
o added citation information
Changes in Version 0.4.4-5.1
o changed internal datapreperation tasks to keep the order of person (estimates) for output.
Changes in Version 0.4.4-4
o extended the main function 'pair()' to use case weights for item parameter calculation.
Changes in Version 0.4.4-3
o changed internal datapreperation tasks (rownames for persons).
o changed: pairwise will globaly change 'options(stringsAsFactors = FALSE)' when applying 'pair()' or 'pers()'
Changes in Version 0.4.4-2
o fixed (changed) summary funtion for objects of class 'pairwise.person.fit'.
o extended function Q() with two arguments to compute the Q person-fit-index based on response data and threshold parameters.
Changes in Version 0.4.4-1
o cleaned up in function Q() to compute the Q person-fit-index proposed by Tarnai and Rost (1990).
Changes in Version 0.4.3-1
o added function Q() to compute the Q person-fit-index proposed by Tarnai and Rost (1990).
Changes in Version 0.4.2-1
o added function q3() to compute Yen's Q3 fit statistic.
o changed structure of result object of function rfa().
Changes in Version 0.4.2-1
o fixed a bug in function ptbis().
Changes in Version 0.4.2
o enhanced plotting method for object of class "pers".
o added function pairwise.S() to perform the Fischer-Scheiblechner statistic 'S'; a 'Wald like test' on item level.
Changes in Version 0.4.1
o added function deltapar() to compute delta or rather item step parameters from thurstonian threshold parameters.
o added argument p in function logLik.pers().
Changes in Version 0.4.0
o fixed minor bug in function pers() relating to passing throug arguments from "pair" object under special data situations.
o fixed minor bug in function andersen.pers() concerning reading arguments from "pers" object.
Changes in Version 0.3.2
o further enhancements.
o fixed a minor bug in function gif().
o fixed a minor bug in plotting methods for item parameters.
Changes in Version 0.3.0
o added wighted Likelihood in result object of function pers().
o added summary function for pairwise.person.fit().
o added summary function for pairwise.item.fit().
o added function simra().
o added function andersentest.pers().
o added covariate data to the bfiN and bfiN_miss data sets.
o added logical option itemNames = 'TRUE / FALSE' in ploting method for objects of class "pers".
o changed structure of result in pairSE() and changed depending functions.
o added the function lrtest.pers() to perform a likelihood ratio test against the saturated model for object of class "pers"
o changed internal calculations for powers of the pairwise comparison matrix.
o make use of transpose logical in S3 method for rfa - plot.rfa().
o extended options of argument pot in func. pair().
o fixed a bug in S3 method logLik.pers when calculating the degrees of freedom.
Changes in Version 0.2.5
o added the S3 method logLik.pers to extract the Log-Likelihood for object of class "pers"
o added the function rfa() and its plotting and summary S3 method to perfom rasch residual factor analysis proposed by Wright (1996) and further discussed by Linacre (1998).
o added the function iff() to plot or return values for Item Information Function.
o added the function tff() to plot or return values for Test Information Function.
o added two example dataset (kft5 and Neoffi) used in Rost (2004).
o added an optional argument in function pers() for calculation of the WLE reliability.
o changed spliting argument in function grm() and changed S3 methods plot and summary for objects of class "grm"
o fixed a minor bug in S3 method plot.grm when defining axis annotation
o fixed a error messages which occured under extreme sparse data situations using function pers()
Changes in Version 0.2.4
o fixed a mismatch when calcualting item and person fit statistic under booklet design circumstances
o fixed some error messages which occur under extreme sparse data situations in some internal functions.
o added WLE Reliability in object of class "pers".
o changed summary method for object of class "pers".
o added plotting method for object of class "pers" and documentation. This plots a person item map.
Changes in Version 0.2.2
o added main function pair() for item parameter calculation for dicho- or polytomuous data . Data with unequal number of categories for items can be processed.
o added main function pers() for person parameter (WLE) calculation for dicho- or polytomuous data. Data with unequal number of categories for items can be processed.
o added function ptbis() for calculation of point biserial correlations for dicho- or polytomous item categories with total scale.
o added person - pairwise.person.fit() - and item - pairwise.item.fit() - fit indices.
o added function for tabulation of response data - ftab().
o added function for expected score curves plots - esc().
o added function for category probability plots - catprob().
o added function for graphical item fit by empirical category probability plots - gif().
o added another (small) example dataset sim200x3, containing simulated Data with one dichotomous and two polytoumous items with 200 observations.
o added a (large) example data structure DEU_PISA2012 - for a detailed description of the data structure see the section User guides, package vignettes and other documentation.
o changed the plotting methods for "pair"", "grm" and "pairSE" and added documentation for them.
o removed deprecated functions (from version 0.1.5):
itempar.dicho(), itempar.poly(), grmSE.dicho(), grmSE.poly(), ipSE.dicho(), ipSE.poly(), ppML.dicho(), ppWL.dicho(), thetaML.dicho(), thetaWL.dicho, ptbis.dicho().
Those functions are therefore only available in older version of pairwise in the CRAN archive.
----------------------------------------------------------------------------------------
Changes in Version 0.1.5
o added function for Point Biserial Correlations ptbis.dicho() (dicotoumous items)
o added function for handling booklet designs make.incidenz()
o added functions for person parameter estimation (ML & WL) and summary method (dicotoumous items)
o added summary method for graphical model-check dichotomuos (object of class "grmSE")
o added summary method for graphical model-check polytomuos (object of class "grmSEpo")
Changes in Version 0.1.4
o added the argument zerocor in the functions for parameter calculations - see descriptions of functions
o changed some matrix calculations to enhance processing speed (due to a tip from Alexander Robitzsch)