First-order and Second-order Equity in Equating
نویسنده
چکیده
iii Brennan Equity The notion of equity in equating was introduced by Lord (1980. pp. 195ff). The basic idea is that for any true score on Form Y, the distribution of observed scores on Form Y should be the same as the distribution of converted scores on Form X (see Kolen & Brennan, 2004. pp. 10–11). If this goal is achieved, Lord argued, then it should be a matter of indifference to each and every examinee which form of a test he or she took. Lord showed, however, that scores on fallible forms of a test cannot be equated under this strict definition of equity. The only exception is when the two forms are parallel in the most strict sense (i.e., indistinguishable forms), in which case equating is unnecessary. So, under Lord's definition of equity, it is sometimes stated that equating is either impossible or unnecessary! Morris (1982) suggested considering a less strict definition of equity, which he called " weak " equity, or first-order equity (FOE), which focuses only on the mean of the distributions of observed scores on Form Y and the converted Form X scores. Today it is rather common to refer to the combination of first-order equity and second-order equity (SOE) as " weak " equity. SOE considers the variance of the distributions of observed scores on Form Y and the converted Form X scores. (We will be more specific about definitions after notation is introduced, below.) This paper considers requirements for attaining, or nearly attaining, FOE and SOE for true-score and observed-score equating, primarily under certain basic assumptions in classical test theory. Linear equating is considered first, followed by curvilinear equating. Particular consideration is given to the role of reliability in attaining, or nearly attaining, FOE and SOE. As discussed more fully later, equal reliability for forms has long been regarded as a requirement for equating, but high reliability has an ambiguous status in the equating literature. An important purpose of this paper is to provide a firmer ground for considering these matters.
منابع مشابه
Dichotomous or polytomous model? equating of testlet-based tests in light of conditional item pair correlations
The performance of dichotomous and polytomous IRT models in equating testletbased tests was compared in this study. To clarify the conditions under which dichotomous and polytomous item response models produce differing results, the DIMTEST program was used for testing essential unidimensionality, and a bias-corrected index (Final Condcorr) was adapted in this study for measuring local item dep...
متن کاملPrioritizing the Factors Affecting Brand Equity of Popular Football Clubs in Iran
Background. Today, Building strong brands, Because of the extraordinary advantages that it creates, have been the marketing priorities of many organizations. Strong brands for a company, Creates an identity in the market. Objectives. This study aimed at Prioritizing factors affecting brand equity of popular football clubs in Iran. Methods. The design of the study was descriptive with practica...
متن کاملThe Missing Data Assumptions of the Nonequivalent Groups With Anchor Test (NEAT) Design and Their Implications for Test Equating
As part of its nonprofit mission, ETS conducts and disseminates the results of research to advance quality and equity in education and assessment for the benefit of ETS's constituents and the field. To obtain a PDF or a print copy of a report, please visit: Abstract The nonequivalent groups with anchor test (NEAT) design involves missing data that are missing by design. Three popular equating m...
متن کاملOn the second order first zagreb index
Inspired by the chemical applications of higher-order connectivity index (or Randic index), we consider here the higher-order first Zagreb index of a molecular graph. In this paper, we study the linear regression analysis of the second order first Zagreb index with the entropy and acentric factor of an octane isomers. The linear model, based on the second order first Zag...
متن کامل