摘要:

The problem of assessing the relative importance of variable subsets in discriminant analysis is discussed. Attention is focused on techniques for determining those subsets which are ‘adequate’ for discrimination from the descriptive viewpoint. A number of procedures which have been proposed or used in the literature are described, illustrated and compared with reference to various aspects, including rationale, statistical significance testing and computational difficult

ISSN:0007-1102    

DOI:10.1111/j.2044-8317.1982.tb00638.x

出版商:Blackwell Publishing Ltd    

年代:1982

数据来源: WILEY

摘要:

The literature dealing with the selection of variables for allocation procedures is examined. Two types of technique are discussed, namely, those which use group separation as the criterion for variable selection and those which more appropriately employ error rates in allocation.

ISSN:0007-1102    

DOI:10.1111/j.2044-8317.1982.tb00639.x

出版商:Blackwell Publishing Ltd    

年代:1982

数据来源: WILEY

摘要:

The test‐taking behaviour of some examinees may be so unusual that their test scores cannot be regarded as appropriate measures of their ability. Appropriateness measurement is a model‐based approach to the problem of identifying these test scores. The intuitions and basic theory supporting appropriateness measurement are presented together with a critical review of earlier work and a series of interrelated experiments. We conclude that appropriateness measurement techniques are robust to errors in parameter estimation and robust to the presence of unidentified aberrant examinees in the test norming sample. In addition, the frequently criticized ‘three‐parameter logistic’ latent trait model was found to be adequate for the detection of spuriously low scores in actual

ISSN:0007-1102    

DOI:10.1111/j.2044-8317.1982.tb00640.x

出版商:Blackwell Publishing Ltd    

年代:1982

数据来源: WILEY

摘要:

Wilcox (1981a) proposed a model for an answer‐until‐correct scoring procedure that solves various measurement problems. The purpose of this paper is to check empirically an implication of the model, and to propose and investigate some strong true‐score models. One of the strong true‐score models assumes the probability of guessing the correct response to an item is a strictly increasing function of an examinee's ability level, and the model gives a reasonable fit to the data. The paper illustrates that this new model is easily applied to situations where the beta‐binomial model is typically used. The other models, including the Dirichlet‐multinomial model, proved to be unsatisfactory. Finally, potential difficulties with the new model are discussed, and possible directions for future research ar

ISSN:0007-1102    

DOI:10.1111/j.2044-8317.1982.tb00641.x

出版商:Blackwell Publishing Ltd    

年代:1982

数据来源: WILEY

摘要:

Theories are put forward that attempt to answer the practical question of ‘How should we correct for guessing in multiple choice tests?’ and the theoretical question of ‘How can we mathematically describe partial knowledge so as to predict behaviour in tasks which enable it to be shown?’. Empirical findings relating to performance on variants of the multiple choice task are reviewed, and compared to the predictions of the t

ISSN:0007-1102    

DOI:10.1111/j.2044-8317.1982.tb00642.x

出版商:Blackwell Publishing Ltd    

年代:1982

数据来源: WILEY

摘要:

Several models exist for the observed numberZof correct answers in a multiple choice test. One of these is the binomial error model which assumes that the conditional distribution ofZgiven the proportion‐correct true score is binomial. The specification of a beta distribution to the proportion‐correct true score has led to the development of the beta‐binomial model. Recently, extensions of the beta‐binomial model have been given that allow for guessing. LettingXbe the number of known items in the test the regression ofXonZhas received much attention. In the present paper it is argued that the numberWof known items in the item pool is of interest rather thanX, and the regression ofWonZis

ISSN:0007-1102    

DOI:10.1111/j.2044-8317.1982.tb00643.x

出版商:Blackwell Publishing Ltd    

年代:1982

数据来源: WILEY

摘要:

Henrysson (1963) has criticized the procedure of correlating an item with the sum of the remainingn− 1 items on the grounds that the criterion is a function of the items selected for analysis. Worst case distortions of the point biserial correlations are examined, and it is concluded that for many data sets such effects are likely to be negligible. In such cases it is argued that the corrected coefficients proposed by Henrysson (1963) should be avoided because of their strong assumptions concerning the factorial homogeneity of the test item

ISSN:0007-1102    

DOI:10.1111/j.2044-8317.1982.tb00644.x

出版商:Blackwell Publishing Ltd    

年代:1982

数据来源: WILEY

摘要:

The average product moment correlation between randomly sampled variate values and their ranks is estimated for small size samples from 10 different distributions. We find that these correlations are generally quite high and approach the limiting values reported by Stuart (1954, 1955) as the sample size increases. These findings have implications for the use of rank‐order values as measures of underlying variate values when the sample size is smal

ISSN:0007-1102    

DOI:10.1111/j.2044-8317.1982.tb00645.x

出版商:Blackwell Publishing Ltd    

年代:1982

数据来源: WILEY

摘要:

A statistical paradigm proposed by Hubertet al.(1981) for evaluating the pattern of cross‐classified proximity data can be used to assess homogeneity between responses for the same subjects on two different tasks, and conversely, for the same tasks over different subjects. The paradigm involves ‘blocking’ on one variable while evaluating the effects of the second. As an example using a similarity rating task and a sorting task on the same animal names, the increase of within subject and within task correlations was conf

ISSN:0007-1102    

DOI:10.1111/j.2044-8317.1982.tb00646.x

出版商:Blackwell Publishing Ltd    

年代:1982

数据来源: WILEY

摘要:

Considerable recent work has been devoted to investigation of the occurrence and effects of letter transpositions within words. Their occurrence has been studied in disorders of reading and writing, and their effects studied experimentally using for example lexical decision tasks. With the publication of empirically compiled tables of anagrammatic frequencies it becomes possible to test a model of the distributions of orthographically defined words of this type. The model rests upon the simplifying assumption that the constraint imposed by the specification of each letter of each verbal item may be taken as fixed.

ISSN:0007-1102    

DOI:10.1111/j.2044-8317.1982.tb00647.x

出版商:Blackwell Publishing Ltd    

年代:1982

数据来源: WILEY