مقایسه نرم افزار بایلوگ با وین استپز

Notes	Bilog	Winsteps
It is said that Rasch modeling is not the same as 1P modeling in Bilog. Rasch is a philosophy of psychometrics, in which data fits the model, not the model fits the data	One-, two-, and three parameter models	Rasch model	Modeling
You can enter scaling options in Winsteps and Bilog.	Default: Probits This is a escaling by 1.7	Default: Logits	Scale unit
Winsteps makes no assumptions about parameter distributions. Bilog assumes normal sample distribution. This may squeeze or spread results particularly at the tails.	Normal sample distribution	No assumption	Assumptions
MMLE assumes the conditional independence of responses to different items by persons of the same ability. UCON is more biased than conditional methods, but this bias is negligibly small and always less than the standard errors of the estimated measures. This usually has only decimal  place effects.	Marginal Maximum Likelihood Estimate as the default. MMAP and Bayes are also available	Joint Maximum Likelihood Estimate, also known as Unconditional Maximum Likelihood estimate (UCON).	Estimation
	Person mean=0 Person variance=1	Item mean=0	Setting of origin
	Multiple-form equating by using common or linking items	Across-sample test equating by using anchored items	Test equating
The numbers in the step function output are difficulty indices in terms of logit, the natural log of the odds ratio. Going from one point to two points, and from two points to three points, will certainly increase the logit difficulty. Distances in logit are comparable. To be specific, if step3 - step2 = 0.1 and step2 - step1 = 0.1. The two "0.1" are considered the same quantities.	Can handle  binary responses only	Can handle both dichotomous and partial credit items, use step functions	Partial credit
	Fit statistic is expressed as Chi-square/degree of freedom, where Chi-square results are testing the fit between the expected and the observed.	Winsteps has two types of fitness  indexes: INMSQ (Infit mean square) and OUTSQ (Outfit mean square). The INMSQ is usually more informative than the OUTSQ.	Fitness

منبع:

Yu, Chong Ho & Sharon E. Osborn Popp (2005). Test Equating by Common Items and Common Subjects: Concepts and Applications. Practical Assessment Research & Evaluation, 10(4). Available online: http://pareonline.net/getvn.asp?v=10&n=4