2021-09-07 Stephane Germain  <germste@gmail.com>

	* Remove the check for ordered modal threshold in the graded model

	* version 1.4.0

2018-08-29 Stephane Germain  <germste@gmail.com>

	* Updated compilation scripts for Windows

	* Windows package compiled with gsl 2.5

	* version 1.3.1

2013-02-21  Stephane Germain  <germste@gmail.com>

	* added the computation of biserial/polyserial correlation in classical.c

	* version 1.2.0

2013-02-19  Stephane Germain  <germste@gmail.com>

	* added the file polyserial.c

2012-06-13  Stephane Germain  <germste@gmail.com>

	* change the calculations of the score mean, item-tot
	correlations, alpha and score sd to work when there is no complete
	observation

	* release version 1.0.3

	* Did I just undo the change from 2011-04-15 !! Some want the
	results of the CTT to be the same as SPSS (ie delete subjects with
	missings) and others want to be able to use it even if they have
	no or very little complete subject (assuming MAR or MCAR). I
	choose the later for now.

2011-04-15  Stephane Germain  <germste@gmail.com>

	* Ignore the incomplete subjects in the classical test theory statistics

	* Test for a singular hessian in the newton iterations

	* release version 1.0.2

2010-11-05  Stephane Germain  <germste@gmail.com>

	* added a stop condition when a NAN is obtained in the EM algo

	* added handling of NAN in the root functions evaluation

	* release version 1.0.1

2010-01-18  Stephane Germain  <germste@gmail.com>

	* release version 1.0.0

2009-11-30  Stephane Germain  <germste@gmail.com>

	* correct a bug in classical_statistics_mc for the computation
	of the correlation matrix when the weight is nominal

2009-11-24  Stephane Germain  <germste@gmail.com>

	* change the way the derivatives are computed in info_from_probs
	and info_from_probs_mc, we now use cubic-spline interpolation

	* add the test info as a separete output parameter in info_from_probs
	and info_from_probs_mc

	* add Warm's weighted maximum likelihood estimator (WMLE) for the
	abilities (wmle_ability.c)

2009-11-23  Stephane Germain  <germste@gmail.com>

	* change the degrees of freedom used in the local independance test
	for the mc models

	* add the df as output parameter in llk_ratio_ld_test
	and llk_ratio_ld_test_mc

	* add the correlation between each item pairs as output in
	classical_statistics and classical_statistics_mc

2009-06-14  Stephane Germain  <germste@gmail.com>

	* fix an error in the test of fit for the multiple-choice, the
	statistic was sometime negative when some options had a zero
	frequency in some class, the effect on the p-value was negligible

2009-04-07  Stephane Germain  <germste@gmail.com>

	* compute the standard errors at each iteration for the mc and
	graded models

2008-01-31  Stephane Germain  <germste@gmail.com>

	* check for the convergence of the likelihood in all the EM

	* reset the prob if too small or too big in prob_3plm

2007-09-17  Stephane Germain  <germste@gmail.com>

	* fix a bug in classical_statistics_mc for the correlations

2007-09-03  Stephane Germain  <germste@gmail.com>

	* change the quad_sizes from a matrix(items x classes) to a
	vector(classes), since it is all the same for all the items

	* adjust the quadrature weights at the end of each EM iteration

2007-08-30  Stephane Germain  <germste@gmail.com>

	* reset the prob if too small or too big in probs_2plm_grad, 
	probs_2plm_mc and like_2plm_mc_dfdf2

	* make sure the boundary thresholds are ordered in mle_2plm_grad

	* fix the thresholds assignement in set_ignore_mc

	* add the compilation notes at the end of the INSTALL file

2007-08-28  Stephane Germain  <germste@gmail.com>

	* reset the prob if too small or too big in like_2plm_grad_dfdf2

2007-07-19  Stephane Germain  <germste@gmail.com>

	* compute the log likelihood ratio test for local dependance

2007-07-18  Stephane Germain  <germste@gmail.com>

	* compute the classical test theory statistics

2007-07-17  Stephane Germain  <germste@gmail.com>

	* compute the log likelihood ratio goodness of fit test
	for multiple choice item

2007-07-13  Stephane Germain  <germste@gmail.com>

	* compute the log likelihood ratio goodness of fit test
	for binary item

2007-07-11  Stephane Germain  <germste@gmail.com>

	* compute the information functions

2007-07-10  Stephane Germain  <germste@gmail.com>

	* compute the icc

2007-07-09  Stephane Germain  <germste@gmail.com>

	* compute the standard errors in mmle_2plm_grad

2007-07-05  Stephane Germain  <germste@gmail.com>

	* add the graded model

2007-07-03  Stephane Germain  <germste@gmail.com>

	* fixed the way the blanks are treated in nadaraya_watson_mc
	and also in nadaraya_watson

2007-07-02  Stephane Germain  <germste@gmail.com>

	* reset the logit bigger than 23 in pmmle_wave_mc

2007-06-28  Stephane Germain  <germste@gmail.com>

	* add the multivariate model

2006-07-10  Stephane Germain  <germste@gmail.com>

	* when an item do not converge leave
	its parameters unchanged in bme_3plm

	* no longer break when the iterate
	function return an error as "no progress"
	in bme_3plm

2006-07-03  Stephane Germain  <germste@gmail.com>

	* reset the probs to be inside the interval (0,1)
	in em_mple_wave

	* added more debugging information at verbosity
	levels 5 and more

	* added a stopping criteria if hte change
	in the PML is too small between EM iteration
	in em_mple_wave

2006-06-30  Stephane Germain  <germste@gmail.com>

	* added support for degenerate items in bme_3plm and
	em_mple_wave

2006-06-29  Stephane Germain  <germste@gmail.com>

	* corrected an error in the computation of
	the log_slope_mean and log_slope_var
	in bme_3plm

2006-04-24  Stephane Germain  <germste@gmail.com>

	* added no_trans in the struct bme_3plm_struct,
	to calculate the Hessian without transformations
	on the slope and asymptote

	* calculate accurate standard error in bme_3plm

	* correct the likelihood, gradient and Hessian
	for missing value in mle_ability_3plm_fdfdf2

2006-04-08  Stephane Germain  <germste@gmail.com>

	* change the default bandwith by a factor of 1.1
	in accordance with testgraf in the main of irt.c

2006-04-05  Stephane Germain  <germste@gmail.com>

	* initial release 0.1
