Change file for package psych

to do:
modify principal so that scoring of components can yield unstandardized scores
correct weird factor score estimates in omega 
allow ICLUST to do confirmatory clustering
add standard errors to factor loadings

Changes in Version 1.0.86 (2010-03-15)
 o	Changed the default behavior of print.psych.fa to have cut=0 (print all loadings)
 o	Added the ability to factor analyze covariance matrices (with resulting loadings of covariances rather than correlations) (suggested by Andreas Moeltner)
 o	Added two new functions to estimate the greatest lower bound (glb.algebraic written by Andreas Moeltner) and glb.communality.  
 o	Added read.clipboard.fwf to allow reading of fixed width files from clipboard
 o	 Added a sim.rasch function to simulate Rasch models
 o	 Minor change to comorbidity to clean up output (using print.psych)
 o	 Seriously modified the estimation of cohen's Kappa in the wkappa function and added a new function: cohen.kappa to calculate kappa from raw data or from similarity matrices.  Also calculates the variances following Fleiss, Cohen, and Blashfield, and then find the confidence limits.
 o	Added the sim.minor function to simulate correlation matrices with nfact major factors and nvar/2 minor factors. 
  o	Bug fix:  Print factor loadings gave the wrong communalities for sorted loadings

Changes in version 1.0.85  (2009-12-15)
 o  Modified the output from fa.parallel to print summary statistics.  Also added an option to use smcs as communality estimates.
 o  Added the sim.minor function to generate major and minor factor structures. 
 o	Modified sim.structural and make.structure to properly handle structure matrices.
 o	Changed fa to use first derivatives based upon the model rather than empirical.  This results in roughly a 10 times improvement in speed for large problems.  As a consequence, VSS is much faster when using minres than previously.
 o	Improved documentation for the dia functions to show the power of structure diagrams
 o	Improved structure.diagram to allow graphs to go left to right (default) or bottom to top (which means causal arrows go left to right)
 o	Added the affect data set which includes two experimental studies of affect 
 o	added some features from pairs to allow group coloring in pairs.panels
 o	Allow fa.parallel to do minres factoring by default
 o	Modified fa to give both the raw eigen values and the eigen values of the Change file for package psych


Changes in version 1.0.84  (2009-11-30)
 o	Improved documentation for the dia functions to show the power of structure diagrams
 o	Improved structure.diagram to allow graphs to go left to right (default) or bottom to top (which means causal arrows go left to right)
 o	Added the affect data set which includes two experimental studies of affect 
 o	added some features from pairs to allow group coloring in pairs.panels
 o	Allow fa.parallel to do minres factoring by default
 o	Modified fa to give both the raw eigen values and the eigen values of the common factor space. 
 o	Modified pairs.panels so it will plot even if some correlation is NA
 o	Added a legend and improved the output of cor.plot.
 o	Fixed a problem with printing of the principal (components) results
 o	Having learned how to use strwidth, have cleaned up the dia.rect, fa.diagram, structure.diagram, and iclust.diagram functions.
 o	Changed the way fa.parallel handles correlation matrices if the number of subjects is not specified. Rather than issuing a warning and proceeding as if n.obs = nvars, it now assumes n.obs =100.
 o	Improved documentation for fa to explain minimum correlation of factor score estimates.  
 o	Fixed bug in the print.psych.vss function that returned the wrong value (although idenitifying the correct number) for the  Velicer MAP criterion. (Reported by Maike Luhmann).
 
Changes in version 1.0.83  (2009-10-26)
 o	Added Suggests Rgraphviz to DESCRIPTION file to get around CRAN test for 2.10 (as I should have done according to documentation for packages).  Removed all tests of functions which require RGraphviz for 2.9.2 because Rgraphviz does not work on SnowLeopard. 
 



Changes in version 1.0.82  (2009-10-26)
  o	Fixed a bug in fa  that produced incorrect estimates for rotated factors if some of them were negatively correlated. (discovered by Erich Studerus)
  o	Modified alpha, ICLUST and guttman to handle the case of just two variables. (A problem pointed out by Erich Studerus.)
  o	alpha will give appropriate statistics, ICLUST and guttman will complain that their statistics are not really meaningful for the case of two variables.  
  o	Identified and fixed a problem with matrix output for multiple groups in describe.by
  o	Improved the print and summary for mat.regress

Changes in version  1.0.81 (2009-10-04) 
  o	Improved the documentation for the dia.X functions and made some of the calls simpler.  

Changes in version  1.0.80 (2009-09-30) 
  o	Replaced fa.graph, omega.graph, ICLUST.rgraph and structure.graph (which required Rgraphviz) with fa.diagram, omega.diagram, iclust.diagram and structure.diagram (which do not)
  o	added fa.sort to sort factor analysis output (used in fa.diagram as well).

Changes in version  1.0.79 (2009-08-25)
  o	modified harmonic.mean and geometric.mean to accept (by default) na.rm=TRUE
  o	Cleaned up describe.by so that it can handle matrix output of multiple grouping variables and then print correctly
  o	corrected fa to get better gls and wls fits 
  o	added lambda6 reliability estimates to cluster.cor
  o	added winsor.var, winsor.sd, and winsor.mean (to supplement winsor.means) per suggestion of Jim Lemon
  o	modified error.bars and error.bars.by to better treat missing and completely missing data (suggestion by Junqian Gordon Xu).

Changes in version  1.0.78  (2009-07-28)
  o	Corrected bug to ICLUST introduced in version 74 (for sign of beta estimate)
  o	added a fix to factor.stats so that it can handle seriously deficient matrices.
  o	added a fix to principal to check case when sum of loadings is exactly 0 and thus the sign of the sum is not defined. This led to a problem in estimating communalities. (problem reported by Mark Leeds)

Changes in version  1.0.77 (2009-7-24)
  o	Corrected the estimation of communalities  and eigen values in print.psych.fa  for the case of oblique rotations.  (The problem with communalities was reported by Tamaki Hattori).
  o	Corrected a bug in error.bars reported by Junqian Gordon Xu.  (Was drawing multiple bars in case of missing data.)
  o	Fixed a problem with weird multiple R in Omega

Changes in version  1.0.76 (2009-7-4)
  o	fixed incorrect p value for negative correlations in r.test
  o	Yet another check for polycor to get through the Mac test on CRAN.

Changes in version  1.0.75 (2009-07-01)
  o	Report (once again) communalities and uniquenesses for principal components and factors (reported missing by Ista Zahn).
  o	Added two more data sets for examples for factor analysis   -- data(Harman).
  o	Added yet another check to see if polycor is available. The Mac version keeps failing because polycor is NOT available for the test suite. 
  o	Although those functions that require polycor fail gracefully if it is not there, the examples do not.

Changes in version  1.0.74 (2009-06-25) 
  o	Revised ICLUST to find weighted beta.  This is a better estimated of the general factor.  (and introduced a bug that was detected and fixed in version 78) 
  o	Modified the examples for ICLUST to show how to get graphics output with using Rgraphviz.
  o	Added fm="ml" option to the fa (factor analysis) function to do maximum likelihood factoring


Changes in version  1.0.73 (2009-06-08) 
  o	Added General Least Squares (gls) to fitting in fa
  o	Revised psych_overview vignette.
  o	Added mat.sort to sort correlation matrices to make pretty pictures.
  o	Starting to revise the estimate for factor indeterminancy to take into account the other factors.  This is clearly exploratory at this point and is not yet part of the package.
  o	Added a check in psych.demo() to check if the polycor package is available.  This gets around a problem in the Mac version at CRAN (I hope.)
  o	Fixed a bug in fa(fm="wls") for 1 factor problems (reported by Ben Schalet)

Changes in version  1.0.72 (2009-06-02)
  o	Fixed a bug introduced earlier today when printing unsorted omega general factor loadings.
  o	Added factor adequacy measures to omega

Changes in version  1.0.71 (2009-06-01)
  o	Reports omega for 1  factor problem
  o	Revised print.psych to be a number of subroutines for easier maintenance.
  o	Added function fa to combine factor.pa, factor.wls, factor.minres into one function.
  o	Added rotation more rotation options to fa.
  o	Fixed bug in print.psych.omega so that item numbers are correctly identified when sorting.

Changes in version  1.0.70 (2009-05-25)
  o	Added factor.stats function to clarify the various statistics for factor analysis using the factor.pa, factor.minres, and principal functions.
  o	Added "cluster" as a rotation option in target.rot. 
  o	Added omega_infinity to the omega function
  o	adjustment to sim.hierarchical to make consistent with Jensen and Weng
corrected bug in principal for case of singular matrices (reported by Gudmundur Arnkelsson) 
  o	corrected bug in print.psych so that it now correctly sorts by loadings if desired

Version 1.0.69 April 25
  o	Modified cosinor to allow for regression as well as optimization approaches

Version 1.0.68  April 15, 2009
  o	Added factor.minres (basically just factor.pa with a new method) to do minimum residual exploratory factor analysis. 
  o	Added the Holzinger.9 nine cognitive variable data set from the LISREL examples.
  o	Added the msq (motivational state questionnaire) data set of 74 mood items and 3800 subjects.
  o	Added sim.anova to simulate ANOVA and lm experimental designs.
  o	Added target.rot to do targeted factor transformations (basically a Procrustian transformation) to an arbitrary target matrix.


Version 1.0.67  March 24, 2009
  o	Have included vignettes in the package
  o	Rewrote mat.regress to allow for raw data/covariances as input
  o	Seriously simplified the mat.regress code
  o	Added reliability G6+ to alpha, score.items, cluster.Cor
  o	Changed the correction for item overlap to use smc and G6+ theory
  o	Added corr.test to give correlations, sample sizes, and probability values for data.frames and matrix input.
  o	Added bar graph output to the error.bars function.  (Although these "dynamite plots" should be discouraged.)
  o	Revised count.pairwise to be much cleaner code. (Replaced 20 lines with 1).

Version 1.0.66   February 12, 2009
  o	Having finally learned how to write vignettes, have started to write them.
  o	Modififed ICLUST to default to show labels on graphics
  o	Modified factor.congruence to allow comparisons for omega solutions
  o	Modified error bars to allow for raw data input.

Version 1.0.65  February 7, 2009
  o	Released to CRAN (basically 1.0.64 with two errors in the Rd.files fixed).
Version 1.0.64  February 2, 2009
  o	Added the data set Thurstone to the bifactor data set.
  o	Minor corrections to omega, structure.graph and structure.sem functions to allow for ease of use as a front end to sem.

Version 1.0.63  January 22, 2009

  o	Added a tutorial for using psych as a front end to sem
  o	Added structure.graph and structure.sem to create sem commands to represent specified structures.
  o	Added structure.list and phi.list to facilitate making structural models.
  o	Added Promax to ease the finding of promax interfactor correlations.
  o	As a consequence, corrected the use of promax in VSS and omega so it actually works.
  o	Added structure.graph to draw sem/regression figures. I am still having a problem with making the X latent variables line up if they are correlated.
  o	correction to principal for case of rotation not being defined
  o	correction to winsor so that it does not sort the returned vector (this would cause problems in the case of winsorizing matrices or data.frames.  It was not a problem for winsor.mean
  o	reorganization of print.psych function to make for a clearer structure

Version 1.0.62    January 2009
  o	(version number jump because of problems getting through windows tests on CRAN)
  o	added confidence intervals to ICC
  o	renamed a number of functions to make it easier
  o	e.g., the sim.xxx family of simulations
          cortest.xxx family of tests of correlations
  o	added super.matrix function
  o	added data sets for bifactor problems

Change log started January, 2009


Prior revisions are not well documented
Source files available in personality-project.org/r/src/contrib   folder


1.0.58  Sept 9, 2008

1.0.53   July 31, 2008

1.0.53   July 11, 2008
In preparation for ECP short course

1.0.50   June 30, 2008


1.0.46   May 31, 2008

1.0.40   Feb 12, 2008


1.0.35  November 22, 2007


1.0.29   August 14, 2007

 1.0.8  
June 15 2006
R version 2.3

1.05   May 15, 2006

1.0.4    May 15, 2006

1.0.2   March 30, 2006

1.0.1  October 30, 2005
R version 2.2




common factor space. 
 o	Modified pairs.panels so it will plot even if some correlation is NA
 o	Added a legend and improved the output of cor.plot.
 o	Fixed a problem with printing of the principal (components) results
 o	Having learned how to use strwidth, have cleaned up the dia.rect, fa.diagram, structure.diagram, and iclust.diagram functions.
 o	Changed the way fa.parallel handles correlation matrices if the number of subjects is not specified. Rather than issuing a warning and proceeding as if n.obs = nvars, it now assumes n.obs =100.
 o	Improved documentation for fa to explain minimum correlation of factor score estimates.  
 o	Fixed bug in the print.psych.vss function that returned the wrong value (although idenitifying the correct number) for the  Velicer MAP criterion. (Reported by Maike Luhmann).
 
Changes in version 1.0.83  (2009-10-26)
 o	Added Suggests Rgraphviz to DESCRIPTION file to get around CRAN test for 2.10 (as I should have done according to documentation for packages).  Removed all tests of functions which require RGraphviz for 2.9.2 because Rgraphviz does not work on SnowLeopard. 
 



Changes in version 1.0.82  (2009-10-26)
  o	Fixed a bug in fa  that produced incorrect estimates for rotated factors if some of them were negatively correlated. (discovered by Erich Studerus)
  o	Modified alpha, ICLUST and guttman to handle the case of just two variables. (A problem pointed out by Erich Studerus.)
  o	alpha will give appropriate statistics, ICLUST and guttman will complain that their statistics are not really meaningful for the case of two variables.  
  o	Identified and fixed a problem with matrix output for multiple groups in describe.by
  o	Improved the print and summary for mat.regress

Changes in version  1.0.81 (2009-10-04) 
  o	Improved the documentation for the dia.X functions and made some of the calls simpler.  

Changes in version  1.0.80 (2009-09-30) 
  o	Replaced fa.graph, omega.graph, ICLUST.rgraph and structure.graph (which required Rgraphviz) with fa.diagram, omega.diagram, iclust.diagram and structure.diagram (which do not)
  o	added fa.sort to sort factor analysis output (used in fa.diagram as well).

Changes in version  1.0.79 (2009-08-25)
  o	modified harmonic.mean and geometric.mean to accept (by default) na.rm=TRUE
  o	Cleaned up describe.by so that it can handle matrix output of multiple grouping variables and then print correctly
  o	corrected fa to get better gls and wls fits 
  o	added lambda6 reliability estimates to cluster.cor
  o	added winsor.var, winsor.sd, and winsor.mean (to supplement winsor.means) per suggestion of Jim Lemon
  o	modified error.bars and error.bars.by to better treat missing and completely missing data (suggestion by Junqian Gordon Xu).

Changes in version  1.0.78  (2009-07-28)
  o	Corrected bug to ICLUST introduced in version 74 (for sign of beta estimate)
  o	added a fix to factor.stats so that it can handle seriously deficient matrices.
  o	added a fix to principal to check case when sum of loadings is exactly 0 and thus the sign of the sum is not defined. This led to a problem in estimating communalities. (problem reported by Mark Leeds)

Changes in version  1.0.77 (2009-7-24)
  o	Corrected the estimation of communalities  and eigen values in print.psych.fa  for the case of oblique rotations.  (The problem with communalities was reported by Tamaki Hattori).
  o	Corrected a bug in error.bars reported by Junqian Gordon Xu.  (Was drawing multiple bars in case of missing data.)
  o	Fixed a problem with weird multiple R in Omega

Changes in version  1.0.76 (2009-7-4)
  o	fixed incorrect p value for negative correlations in r.test
  o	Yet another check for polycor to get through the Mac test on CRAN.

Changes in version  1.0.75 (2009-07-01)
  o	Report (once again) communalities and uniquenesses for principal components and factors (reported missing by Ista Zahn).
  o	Added two more data sets for examples for factor analysis   -- data(Harman).
  o	Added yet another check to see if polycor is available. The Mac version keeps failing because polycor is NOT available for the test suite. 
  o	Although those functions that require polycor fail gracefully if it is not there, the examples do not.

Changes in version  1.0.74 (2009-06-25) 
  o	Revised ICLUST to find weighted beta.  This is a better estimated of the general factor.  (and introduced a bug that was detected and fixed in version 78) 
  o	Modified the examples for ICLUST to show how to get graphics output with using Rgraphviz.
  o	Added fm="ml" option to the fa (factor analysis) function to do maximum likelihood factoring


Changes in version  1.0.73 (2009-06-08) 
  o	Added General Least Squares (gls) to fitting in fa
  o	Revised psych_overview vignette.
  o	Added mat.sort to sort correlation matrices to make pretty pictures.
  o	Starting to revise the estimate for factor indeterminancy to take into account the other factors.  This is clearly exploratory at this point and is not yet part of the package.
  o	Added a check in psych.demo() to check if the polycor package is available.  This gets around a problem in the Mac version at CRAN (I hope.)
  o	Fixed a bug in fa(fm="wls") for 1 factor problems (reported by Ben Schalet)

Changes in version  1.0.72 (2009-06-02)
  o	Fixed a bug introduced earlier today when printing unsorted omega general factor loadings.
  o	Added factor adequacy measures to omega

Changes in version  1.0.71 (2009-06-01)
  o	Reports omega for 1  factor problem
  o	Revised print.psych to be a number of subroutines for easier maintenance.
  o	Added function fa to combine factor.pa, factor.wls, factor.minres into one function.
  o	Added rotation more rotation options to fa.
  o	Fixed bug in print.psych.omega so that item numbers are correctly identified when sorting.

Changes in version  1.0.70 (2009-05-25)
  o	Added factor.stats function to clarify the various statistics for factor analysis using the factor.pa, factor.minres, and principal functions.
  o	Added "cluster" as a rotation option in target.rot. 
  o	Added omega_infinity to the omega function
  o	adjustment to sim.hierarchical to make consistent with Jensen and Weng
corrected bug in principal for case of singular matrices (reported by Gudmundur Arnkelsson) 
  o	corrected bug in print.psych so that it now correctly sorts by loadings if desired

Version 1.0.69 April 25
  o	Modified cosinor to allow for regression as well as optimization approaches

Version 1.0.68  April 15, 2009
  o	Added factor.minres (basically just factor.pa with a new method) to do minimum residual exploratory factor analysis. 
  o	Added the Holzinger.9 nine cognitive variable data set from the LISREL examples.
  o	Added the msq (motivational state questionnaire) data set of 74 mood items and 3800 subjects.
  o	Added sim.anova to simulate ANOVA and lm experimental designs.
  o	Added target.rot to do targeted factor transformations (basically a Procrustian transformation) to an arbitrary target matrix.


Version 1.0.67  March 24, 2009
  o	Have included vignettes in the package
  o	Rewrote mat.regress to allow for raw data/covariances as input
  o	Seriously simplified the mat.regress code
  o	Added reliability G6+ to alpha, score.items, cluster.Cor
  o	Changed the correction for item overlap to use smc and G6+ theory
  o	Added corr.test to give correlations, sample sizes, and probability values for data.frames and matrix input.
  o	Added bar graph output to the error.bars function.  (Although these "dynamite plots" should be discouraged.)
  o	Revised count.pairwise to be much cleaner code. (Replaced 20 lines with 1).

Version 1.0.66   February 12, 2009
  o	Having finally learned how to write vignettes, have started to write them.
  o	Modififed ICLUST to default to show labels on graphics
  o	Modified factor.congruence to allow comparisons for omega solutions
  o	Modified error bars to allow for raw data input.

Version 1.0.65  February 7, 2009
  o	Released to CRAN (basically 1.0.64 with two errors in the Rd.files fixed).
Version 1.0.64  February 2, 2009
  o	Added the data set Thurstone to the bifactor data set.
  o	Minor corrections to omega, structure.graph and structure.sem functions to allow for ease of use as a front end to sem.

Version 1.0.63  January 22, 2009

  o	Added a tutorial for using psych as a front end to sem
  o	Added structure.graph and structure.sem to create sem commands to represent specified structures.
  o	Added structure.list and phi.list to facilitate making structural models.
  o	Added Promax to ease the finding of promax interfactor correlations.
  o	As a consequence, corrected the use of promax in VSS and omega so it actually works.
  o	Added structure.graph to draw sem/regression figures. I am still having a problem with making the X latent variables line up if they are correlated.
  o	correction to principal for case of rotation not being defined
  o	correction to winsor so that it does not sort the returned vector (this would cause problems in the case of winsorizing matrices or data.frames.  It was not a problem for winsor.mean
  o	reorganization of print.psych function to make for a clearer structure

Version 1.0.62    January 2009
  o	(version number jump because of problems getting through windows tests on CRAN)
  o	added confidence intervals to ICC
  o	renamed a number of functions to make it easier
  o	e.g., the sim.xxx family of simulations
          cortest.xxx family of tests of correlations
  o	added super.matrix function
  o	added data sets for bifactor problems

Change log started January, 2009


Prior revisions are not well documented
Source files available in personality-project.org/r/src/contrib   folder


1.0.58  Sept 9, 2008

1.0.53   July 31, 2008

1.0.53   July 11, 2008
In preparation for ECP short course

1.0.50   June 30, 2008


1.0.46   May 31, 2008

1.0.40   Feb 12, 2008


1.0.35  November 22, 2007


1.0.29   August 14, 2007

 1.0.8  
June 15 2006
R version 2.3

1.05   May 15, 2006

1.0.4    May 15, 2006

1.0.2   March 30, 2006

1.0.1  October 30, 2005
R version 2.2




