|
1. |
Modelling human memory: Connectionism and convolution |
|
British Journal of Mathematical and Statistical Psychology,
Volume 49,
Issue 1,
1996,
Page 1-24
Gordon D. A. Brown,
Charles Hulme,
Peter Dalloz,
Preview
|
PDF (1313KB)
|
|
摘要:
The mathematical operation ofconvolutionis used as an associative mechanism by several recent influential models of human memory. Convolution can be used to associate two vectors (representing items to be remembered) into a memory trace vector in one operation. An approximation to either of the input vectors can then be retrieved, using the other vector as a probe. Recent convolution‐based memory models have accounted for a wide range of data. Connectionist models may have greater potential for providing developmental accounts, but the architectures that have been most widely used to account for developmental phenomena cannot perform one‐trial learning and this has limited their use as models of human memory. We show that a connectionist‐like architecture can learn, using a gradient‐descent algorithm, to perform single‐trial learning in a similar manner to convolution. The solution that the network finds leads to less variable retrieval than does convolution. Furthermore, the network can learn to carry out the convolution operation itself. This provides a link between connectionist and convolution approaches, and a basis for models with many of the attractions of both connectionist and convolution a
ISSN:0007-1102
DOI:10.1111/j.2044-8317.1996.tb01072.x
出版商:Blackwell Publishing Ltd
年代:1996
数据来源: WILEY
|
2. |
Statistical methods for understanding cognitive growth: A review, a synthesis and an application |
|
British Journal of Mathematical and Statistical Psychology,
Volume 49,
Issue 1,
1996,
Page 25-42
Ian Plewis,
Preview
|
PDF (1071KB)
|
|
摘要:
Methods for analysing data on cognitive growth are reviewed with emphasis given to the problems posed by scales which inevitably change over age. Multilevel growth curve and regression models are contrasted. A hybrid model is proposed which shares some of the advantages of these two approaches. The models are applied to a set of longitudinal data on mathematics growth from London primary school pupils. We find that conclusions about relative growth vary according to the scale adopted. Implications for the design and analysis of growth studies in psychology and education are discussed.
ISSN:0007-1102
DOI:10.1111/j.2044-8317.1996.tb01073.x
出版商:Blackwell Publishing Ltd
年代:1996
数据来源: WILEY
|
3. |
A developmental path model and causal analysis of latent dichotomous variables |
|
British Journal of Mathematical and Statistical Psychology,
Volume 49,
Issue 1,
1996,
Page 43-56
Nobuoki Eshima,
Chooichiro Asano,
Minoru Tabata,
Preview
|
PDF (496KB)
|
|
摘要:
In a developmental study, it is reasonable to assume that a population is divided into several subpopulations according to developmental paths, i.e. each of the subpopulations depends on a developmental path of its own. Under this assumption, first, a latent structure model is proposed for explaining developmental paths in a population. Second, causal analysis of latent dichotomous variables indicating states of skill acquisition is discussed through the developmental paths. Basic measures of causal effects among the variables are defined. A numerical example illustrates the present approach.
ISSN:0007-1102
DOI:10.1111/j.2044-8317.1996.tb01074.x
出版商:Blackwell Publishing Ltd
年代:1996
数据来源: WILEY
|
4. |
Analysing multiattribute ranking data: Joint and conditional approaches |
|
British Journal of Mathematical and Statistical Psychology,
Volume 49,
Issue 1,
1996,
Page 57-78
Ulf Böckenholt,
Preview
|
PDF (1076KB)
|
|
摘要:
Rankings, in contrast to ratings, eliminate effects of individual differences in scale usage and avoid arbitrary definitions regarding the number of response categories and category labels. However, despite the appeal and popularity of this technique, few methods are available for the analysis of rankings on several attributes. This paper presents extensions of Thurstonian and logistic models for the joint analysis of multiattribute ranking responses and for conditional analyses where rankings on one of the attributes are modelled as a function of the rankings on the other attributes. These extensions are based on two approaches proposed to account for associations among the ranking responses. Empirical applications of the Thurstonian and logistic ranking models indicate that one of the two approaches appears particularly promising for the analysis of multiattribute ranking data.
ISSN:0007-1102
DOI:10.1111/j.2044-8317.1996.tb01075.x
出版商:Blackwell Publishing Ltd
年代:1996
数据来源: WILEY
|
5. |
A survey of theory and methods of invariant item ordering |
|
British Journal of Mathematical and Statistical Psychology,
Volume 49,
Issue 1,
1996,
Page 79-105
Klaas Sijtsma,
Brian W. Junker,
Preview
|
PDF (1507KB)
|
|
摘要:
In many testing situations, ordering the items by difficulty is helpful in analysing the testing data; examples include intelligence testing, analysis of differential item functioning, person‐fit analysis, and exploring hypotheses about the order in which cognitive operations are acquired by children. In each situation, interpretation and analysis are made easier if the items are ordered by difficulty in the same way for every individual taking the test, i.e. the item response functions do not cross. This is aninvariant item ordering.In this paper we review a class of non‐parametric unidimensional item response models in which the ordinal properties of items (and persons) can be studied, and survey both old and new methods for the investigation of invariant item ordering in empirical data sets. Our model formulation derives in particular fiom the work of Holland&Rosenbaum (1986), Junker (1993) and Mokken (1971). We survey methods based on the work of Mokken (1 97 I), Rosenbaum (1 987a, b), and Sijtsma&Meijer (1992), and we also discuss some new proposals for checking invariant item ordering. When violations are detected, these methods allow a rough assessment of where on the latent scale the item response functions cross.We also study similarities and differences between these various methods and provide guidelines for their use. Finally, the methods are illustrated with data from a developmental psychology experiment in which the ability to draw inferences about transitive relations is explo
ISSN:0007-1102
DOI:10.1111/j.2044-8317.1996.tb01076.x
出版商:Blackwell Publishing Ltd
年代:1996
数据来源: WILEY
|
6. |
Linear composites in multiattribute judgement and choice: Extensions of Wilks' results to zero and negatively correlated attributes |
|
British Journal of Mathematical and Statistical Psychology,
Volume 49,
Issue 1,
1996,
Page 107-126
Abba M. Krieger,
Paul E. Green,
Preview
|
PDF (944KB)
|
|
摘要:
One of the most common ways of evaluating two or more options that vary along multiple criteria is to form linear composites that incorporate the decision maker's assessment of the relative importance of each criterion. Variations of this procedure are prevalent in both the behavioural and policy disciplines.Over 50 years ago Wilks (1 938) showed that if the criteria underlying the options are sufficiently highly correlated then the choice of weights does not matter, in the sense that the correlation between composites formed from any two independent sets of ratings tends to 1 as the number of criteria increases. Wilks' result, however, requires two assumptions, namely positively correlated attributes and independent choices of weights. Through various extensions of his results, we show that Wilks' conclusion is highly robust‐nly extreme, simultaneous violation of both assumptions produces negatively correlated composite
ISSN:0007-1102
DOI:10.1111/j.2044-8317.1996.tb01077.x
出版商:Blackwell Publishing Ltd
年代:1996
数据来源: WILEY
|
7. |
Models for asymmetric proximities |
|
British Journal of Mathematical and Statistical Psychology,
Volume 49,
Issue 1,
1996,
Page 127-146
Bertie Zielman,
Willem J. Heiser,
Preview
|
PDF (1105KB)
|
|
摘要:
All major models for asymmetric proximities are a combination of a symmetric similarity component and an asymmetric dominance component. The differences and similarities between the methods that are discussed in this paper are revealed by applying a certain decomposition to the model parameters, clearly separating the dominance and symmetric similarity component. The notion of skew‐symmetry turns out to be an often seen element in modelling asymmetry, although sometimes in disguise and difficult to recognize. The decomposition shows that there are two classes of models, one that assumes that the asymmetric relationships are transitive, while the other class consists of models that can also represent circular asymmetric relationship
ISSN:0007-1102
DOI:10.1111/j.2044-8317.1996.tb01078.x
出版商:Blackwell Publishing Ltd
年代:1996
数据来源: WILEY
|
8. |
Interaction contrasts in repeated measures designs |
|
British Journal of Mathematical and Statistical Psychology,
Volume 49,
Issue 1,
1996,
Page 147-162
Lisa M. Lix,
H. J. Keselman,
Preview
|
PDF (952KB)
|
|
摘要:
Specific information concerning the nature of interaction effects in factorial designs may be obtained through the use of tetrad contrasts. Empirical familywise Type I error rates and power rates associated with 10 procedures for conducting tetrad contrasts in groups‐by‐trials repeated measures designs were obtained when the assumptions of multisample sphericity and multivariate normality were not satisfied. Only three procedures provided acceptable control of familywise Type I error rates under the assumption violation conditions. These procedures relied on a test statistic formed using an estimate of the standard error of the contrast which was based on only that data used in defining the contrast, in combination with either a Studentized maximum modulus, Hochberg (1988) step‐up Bonferroni, or Shaffer (1986) modified sequentially rejective Bonferroni critical value. Minimal power differences between these three procedures were obs
ISSN:0007-1102
DOI:10.1111/j.2044-8317.1996.tb01079.x
出版商:Blackwell Publishing Ltd
年代:1996
数据来源: WILEY
|
9. |
Confidence intervals for two robust regression lines with a heteroscedastic error term |
|
British Journal of Mathematical and Statistical Psychology,
Volume 49,
Issue 1,
1996,
Page 163-170
Rand R. Wilcox,
Preview
|
PDF (424KB)
|
|
摘要:
Recently, Wilcox (1994) suggested a method of computing a confidence interval for the slope of two robust regression methods. Results from Wu (1986) suggest that Wilcox's method will fail when the error term has non‐constant variance and simulations reported here confirm this. This paper describes a more satisfactory method for handling non‐constant variance when working with the biweight midregression line considered by Wilcox, with little or no loss in accuracy when the error term has constant variance. Moreover, this paper illustrates that the biweight midregression method, in its present form, can yield an unsatisfactory estimate of the slope when there is a heteroscedastic error term, even with a very large sample size. A simple method of correcting this problem is described which also eliminates the need for the bias correction term used by Wilcox. This paper also considers the problem of computing confidence intervals for the slope of the robust regression method derived by De Jongh, De Wet and Welsh (1988), and a fairly accurate method is found even when the error term has non‐constant var
ISSN:0007-1102
DOI:10.1111/j.2044-8317.1996.tb01080.x
出版商:Blackwell Publishing Ltd
年代:1996
数据来源: WILEY
|
10. |
The scaling of dispersion and correlation: A comparison of least‐squares and absolute‐deviation statistics |
|
British Journal of Mathematical and Statistical Psychology,
Volume 49,
Issue 1,
1996,
Page 171-188
Denys Catanzaro,
James C. Taylor,
Preview
|
PDF (909KB)
|
|
摘要:
In conventional measures of dispersion and correlation, the use of squared deviations from the mean causes a disproportionate influence of outliers. We have compared various conventional and unconventional measures of dispersion and correlation through repeated computer trials, taking large numbers of samples from populations of specified distributions. Under the normal distribution, the standard deviation and Pearson correlation provide the most stable estimates of dispersion and correlation. However, in other cases involving outliers within normal distributions, double exponential distributions, or skewed distributions, measures of dispersion and correlation involving absolute deviations from the mean are clearly more stable than arc conventional formulae.
ISSN:0007-1102
DOI:10.1111/j.2044-8317.1996.tb01081.x
出版商:Blackwell Publishing Ltd
年代:1996
数据来源: WILEY
|
|