Carnap's inductive probabilities as a contribution to decision theory

Common probability theories only allow the deduction of probabilities by using previously known or presupposed probabilities. They do not, however, allow the derivation of probabilities from observed data alone. The question thus arises as to how probabilities in the empirical sciences, especially in medicine, may be arrived at. Carnap hoped to be able to answer this question byhis theory of inductive probabilities. In the first four sections of the present paper the above mentioned problem is discussed in general. After a short presentation of Carnap's theory it is then shown that this theory cannot claim validity for arbitrary random processes. It is suggested that the theory be only applied to binomial and multinomial experiments. By application of de Finetti's theorem Carnap's inductive probabilities are interpreted as consecutive probabilities of the Bayesian kind. Through the introduction of a new axiom the decision parameter λ can be determined even if no a priori knowledge is given. Finally, it is demonstrated that the fundamental problem of Wald's decision theory, i.e., the determination of a plausible criterion where no a priori knowledge is available, can be solved for the cases of binomial and multinomial experiments.     

HMMTree: A computer program for latent-class hierarchical multinomial processing tree models

Latent-class hierarchical multinomial models are an important extension of the widely used family of multinomial processing tree models, in that they allow for testing the parameter homogeneity assumption and provide a framework for modeling parameter heterogeneity. In this article, the computer program HMMTree is introduced as a means of implementing latent-class hierarchical multinomial processing tree models. HMMTree computes parameter estimates, confidence intervals, and goodness-of-fit statistics for such models, as well as the Fisher information, expected category means and variances, and posterior probabilities for class membership. A brief guide to using the program is provided.     

Using the conditional grade-of-membership model to assess judgment accuracy

Consider the case whereJ instruments are used to classify each ofI objects relative toK nominal categories. The conditional grade-of-membership (GoM) model provides a method of estimating the classification probabilities of each instrument (or judge) when the objects being classified consist of both pure types that lie exclusively in one ofK nominal categories, and mixtures that lie in more than one category. Classification probabilities are identifiable whenever the sample of GoM vectors includes pure types from each category. When additional, relatively mild, assumptions are made about judgment accuracy, the identifiable correct classification probabilities are the greatest lower bounds among all solutions that might correspond to the observed multinomial process, even when the unobserved GoM vectors do not include pure types from each category. Estimation using the conditional GoM model is illustrated on a simulated data set. Further simulations show that the estimates of the classification probabilities are relatively accurate, even when the sample contains only a small percentage of approximately pure objects.The authors thank Max A. Woodbury, Kenneth G. Manton and H. Dennis Tolley for their help and four anonymous Psychometrika reviewers (including an associate editor) for their beneficial expository and technical suggestions. This work was supported by the Dean's Fund for Summer Research, Owen Graduate School of Management, Vanderbilt University.     

Constructing the exact significance level for a person fit statistic

This paper uses an extension of the network algorithm originally introduced by Mehta and Patel to construct exact tail probabilities for testing the general hypothesis that item responses are distributed according to the Rasch model. By assuming that item difficulties are known, the algorithm is applicable to the statistical tests either given the maximum likelihood ability estimate or conditioned on the total score. A simulation study indicates that the network algorithm is an efficient tool for computing the significance level of a person fit statistic based on test lengths of 30 items or less.     

The statistical analysis of general processing tree models with the EM algorithm

Multinomial processing tree models assume that an observed behavior category can arise from one or more processing sequences represented as branches in a tree. These models form a subclass of parametric, multinomial models, and they provide a substantively motivated alternative to loglinear models. We consider the usual case where branch probabilities are products of nonnegative integer powers in the parameters, 0_s1, and their complements, 1 - _s. A version of the EM algorithm is constructed that has very strong properties. First, the E-step and the M-step are both analytic and computationally easy; therefore, a fast PC program can be constructed for obtaining MLEs for large numbers of parameters. Second, a closed form expression for the observed Fisher information matrix is obtained for the entire class. Third, it is proved that the algorithm necessarily converges to a local maximum, and this is a stronger result than for the exponential family as a whole. Fourth, we show how the algorithm can handle quite general hypothesis tests concerning restrictions on the model parameters. Fifth, we extend the algorithm to handle the Read and Cressie power divergence family of goodness-of-fit statistics. The paper includes an example to illustrate some of these results.This research was supported by National Science Foundation Grant BNS-8910552 to William H. Batchelder and David M. Riefer. We are grateful to David Riefer for his useful comments, and to the Institute for Mathematical Behavior Sciences for its support.     

Analyses of multinomial mixture distributions: new tests for stochastic models of cognition and action

Mixture distributions are formed from a weighted linear combination of 2 or more underlying basis distributions [g(x) = sigma j alpha j fj(x); sigma alpha j = 1]. They arise frequently in stochastic models of perception, cognition, and action in which a finite number of discrete internal states are entered probabilistically over a series of trials. This article reviews various distributional properties that have been examined to test for the presence of mixture distributions. A new multinomial maximum likelihood mixture (MMLM) analysis is discussed for estimating the mixing probabilities alpha j and the basis distributions fj(x) of a hypothesized mixture distribution. The analysis also generates a maximum likelihood goodness-of-fit statistic for testing various mixture hypotheses. Stochastic computer simulations characterize the statistical power of such tests under representative conditions. Two empirical studies of mental processes hypothesized to involve mixture distributions are summarized to illustrate applications of the MMLM analysis.     

The EM algorithm for latent class analysis with equality constraints

The EM algorithm is a popular iterative method for estimating parameters in the latent class model where at each step the unknown parameters can be estimated simply as weighted sums of some latent proportions. The algorithm may also be used when some parameters are constrained to equal given constants or each other. It is shown that in the general case with equality constraints, the EM algorithm is not simple to apply because a nonlinear equation has to be solved. This problem arises, mainly, when equality constrints are defined over probabilities indifferent combinations of variables and latent classes. A simple condition is given in which, although probabilities in different variable-latent class combinations are constrained to be equal, the EM algorithm is still simple to apply.The authors are grateful to the Editor and the anonymous reviewers for their helpful comments on an earlier draft of this paper. C. C. Clogg and R. Luijkx are also acknowledged for verifying our results with their computer programs MLLSA and LCAG, respectively.     

Process-dissociation procedure: a testable model for considering assumptions about the stochastic relation between consciously controlled and automatic processes

This paper presents an extension of the process-dissociation procedure with wordstem completion, which makes possible the measurement of the stochastic relationship between consciously controlled and automatic processes. By means of an indirect wordstem completion test, the conditional probabilities of conscious remembering with and without automatic processes can be successfully determined. A multinomial model for the evaluation of this extended process-dissociation procedure is presented. This model makes the distinction between voluntary and involuntary conscious memory processes possible and has been applied to two experiments discussed in this paper. The results show that the assumption of stochastic independence is often violated, albeit not as strongly as predicted by the redundancy or exclusivity model variants. Two conscious processes were found, voluntary and involuntary conscious memory processes, each with a different probability of occurrence.     

Pooling data versus averaging model fits for some prototypical multinomial processing tree models

For a study with multinomial data where there are n_g individuals and with each person having n_r test trials, the question arises as to how to fit the parameters of a multinomial processing tree (MPT) model. Should each parameter be estimated for each individual and then averaged to obtain a group estimate, or should the frequencies in the multinomial categories be pooled so that the model is fit once for the entire group? This basic question is explored with a series of Monte Carlo simulations for some prototypical MPT models. There is a general finding of a pooling advantage for the case where there is a single experimental condition. Also when there are different experimental conditions, there is reduced bias for detecting condition differences for a method based on the pooled data. Although the focus of the paper is on multinomial models, a general theorem is advanced that establishes a basic condition that determines whether there is or is not a difference between the averaging of individual estimates and the estimate based on the pooled data.     

A high-speed algorithm for computing conditional probabilities of substrings of sequentially observed data

An algorithm is described that computes relative frequencies of occurrence of all arbitrarily long substrings of sequential data, such as are obtained from experiments in learning/memory and verbal interaction. The algorithm offers high speed and provides systematization for the computation of empirical conditional probabilities. Use of this algorithm allows application of probabilistic and information theoretic disciplines to reveal dependencies between events separated arbitrarily in time.     

Reconstructing Distances among Objects from Their Discriminability

We describe a principled way of imposing a metric representing dissimilarities on any discrete set of stimuli (symbols, handwritings, consumer products, X-ray films, etc.), given the probabilities with which they are discriminated from each other by a perceiving system, such as an organism, person, group of experts, neuronal structure, technical device, or even an abstract computational algorithm. In this procedure one does not have to assume that discrimination probabilities are monotonically related to distances, or that the distances belong to a predefined class of metrics, such as Minkowski. Discrimination probabilities do not have to be symmetric, the probability of discriminating an object from itself need not be a constant, and discrimination probabilities are allowed to be 0’s and 1’s. The only requirement that has to be satisfied is Regular Minimality, a principle we consider the defining property of discrimination: for ordered stimulus pairs (a,b), b is least frequently discriminated from a if and only if a is least frequently discriminated from b. Regular Minimality generalizes one of the weak consequences of the assumption that discrimination probabilities are monotonically related to distances: the probability of discriminating a from a should be less than that of discriminating a from any other object. This special form of Regular Minimality also underlies such traditional analyses of discrimination probabilities as Multidimensional Scaling and Cluster Analysis. This research was supported by the NSF grant SES 0318010 (E.D.), Humboldt Research Award (E.D.), Humboldt Foundation grant DEU/1038348 (H.C. & E.D.), and DFG grant Co 94/5 (H.C.).     

Generalized latent variable models with non‐linear effects

Until recently, item response models such as the factor analysis model for metric responses, the two‐parameter logistic model for binary responses and the multinomial model for nominal responses considered only the main effects of latent variables without allowing for interaction or polynomial latent variable effects. However, non‐linear relationships among the latent variables might be necessary in real applications. Methods for fitting models with non‐linear latent terms have been developed mainly under the structural equation modelling approach. In this paper, we consider a latent variable model framework for mixed responses (metric and categorical) that allows inclusion of both non‐linear latent and covariate effects. The model parameters are estimated using full maximum likelihood based on a hybrid integration–maximization algorithm. Finally, a method for obtaining factor scores based on multiple imputation is proposed here for the non‐linear model.     

Emotionally enhanced memory for negatively arousing words: storage or retrieval advantage?

People typically remember emotionally negative words better than neutral words. Two experiments are reported that investigate whether emotionally enhanced memory (EEM) for negatively arousing words is based on a storage or retrieval advantage. Participants studied non-word–word pairs that either involved negatively arousing or neutral target words. Memory for these target words was tested by means of a recognition test and a cued-recall test. Data were analysed with a multinomial model that allows the disentanglement of storage and retrieval processes in the present recognition-then-cued-recall paradigm. In both experiments the multinomial analyses revealed no storage differences between negatively arousing and neutral words but a clear retrieval advantage for negatively arousing words in the cued-recall test. These findings suggest that EEM for negatively arousing words is driven by associative processes.     

Hierarchical Multinomial Processing Tree Models: A Latent-Class Approach

Multinomial processing tree models are widely used in many areas of psychology. Their application relies on the assumption of parameter homogeneity, that is, on the assumption that participants do not differ in their parameter values. Tests for parameter homogeneity are proposed that can be routinely used as part of multinomial model analyses to defend the assumption. If parameter homogeneity is found to be violated, a new family of models, termed latent-class multinomial processing tree models, can be applied that accommodates parameter heterogeneity and correlated parameters, yet preserves most of the advantages of the traditional multinomial method. Estimation, goodness-of-fit tests, and tests of other hypotheses of interest are considered for the new family of models. The author thanks Bill Batchelder, Edgar Erdfelder, Thorsten Meiser, and Christoph Stahl for helpful comments on a previous version of this paper. The author is also grateful to Edgar Erdfelder for making available the data set analyzed in this paper.     

Assessing individual differences in categorical data

In cognitive modeling, data are often categorical observations taken over participants and items. Usually subsets of these observations are pooled and analyzed by a cognitive model assuming the category counts come from a multinomial distribution with the same model parameters underlying all observations. It is well known that if there are individual differences in participants and/or items, a model analysis of the pooled data may be quite misleading, and in such cases it may be appropriate to augment the cognitive model with parametric random effects assumptions. On the other hand, if random effects are incorporated into a cognitive model that is not needed, the resulting model may be more flexible than the multinomial model that assumes no heterogeneity, and this may lead to overfitting. This article presents Monte Carlo statistical tests for directly detecting individual participant and/or item heterogeneity that depend only on the data structure itself. These tests are based on the fact that heterogeneity in participants and/or items results in overdispersion of certain category count statistics. It is argued that the methods developed in the article should be applied to any set of participant 3 item categorical data prior to cognitive model-based analyses.     

On belief bias in syllogistic reasoning

A multinomial model is used to disentangle the respective contributions of reasoning processes and response bias in conclusion-acceptance data that exhibit belief bias. A model-based meta-analysis of 22 studies reveals that such data are structurally too sparse to allow discrimination of different accounts of belief bias. Four experiments are conducted to obtain richer data, allowing deeper tests through the use of the multinomial model. None of the current accounts of belief bias is consistent with the complex pattern of results. A new theory of belief bias is proposed that assumes that most reasoners construct only one mental model representing the premises as well as the conclusion or, in the case of an unbelievable conclusion, its logical negation. New predictions derived from the theory are confirmed in 4 additional studies.     

Pitches that Wire Together Fire Together: Scale Degree Associations Across Time Predict Melodic Expectations

The ongoing generation of expectations is fundamental to listeners’ experience of music, but research into types of statistical information that listeners extract from musical melodies has tended to emphasize transition probabilities and n-grams, with limited consideration given to other types of statistical learning that may be relevant. Temporal associations between scale degrees represent a different type of information present in musical melodies that can be learned from musical corpora using expectation networks, a computationally simple method based on activation and decay. Expectation networks infer the expectation of encountering one scale degree followed in the near (but not necessarily immediate) future by another given scale degree, with previous work suggesting that scale degree associations learned by expectation networks better predict listener ratings of pitch similarity than transition probabilities. The current work outlines how these learned scale degree associations can be combined to predict melodic continuations and tests the resulting predictions on a dataset of listener responses to a musical cloze task previously used to compare two other models of melodic expectation, a variable-order Markov model (IDyOM) and Temperley's music-theoretically motivated model. Under multinomial logistic regression, all three models explain significant unique variance in human melodic expectations, with coefficient estimates highest for expectation networks. These results suggest that generalized scale degree associations informed by both adjacent and nonadjacent relationships between melodic notes influence listeners’ melodic predictions above and beyond n-gram context, highlighting the need to consider a broader range of statistical learning processes that may underlie listeners’ expectations for upcoming musical events.     

Fertility-limiting behavior and contraceptive choice among men in Nepal

CONTEXT: Contraceptive choices among men who want no more children have been little explored in South Asia, particularly in Nepal, where fertility rates have remained high over the last few decades. METHODS: Using the 2001 Nepal Demographic and Health Survey couple data set, multinomial logistic regression analyses were conducted for 1,041 married men aged 20 or older who had at least one living child and wanted no more children. Regression models examined relationships between selected characteristics and men's reported contraceptive use, and predicted probabilities were estimated to assess interactions between ecological zone, family composition and method choice. The primary goal was to determine whether the number and sex of living children influenced contraceptive use. RESULTS: Twenty-four percent of men who wanted no more children were not using any contraceptive method at the time of the survey, 30% reported that their wives were sterilized, 12% had had a vasectomy, 7% were using condoms and 27% used other temporary methods. The probability of relying on permanent methods was highest among men who had at least two living sons and lowest among those who had only daughters, while the probability of using no method was highest among those who had only daughters. CONCLUSION: In Nepal, men who report a desire to have no more children are likely to choose permanent methods only after they have two living sons.     

Bayesian Model Comparison for the Order Restricted RC Association Model

Association models constitute an attractive alternative to the usual log-linear models for modeling the dependence between classification variables. They impose special structure on the underlying association by assigning scores on the levels of each classification variable, which can be fixed or parametric. Under the general row-column (RC) association model, both row and column scores are unknown parameters without any restriction concerning their ordinality. However, when the classification variables are ordinal, order restrictions on the scores arise naturally. Under such restrictions, we adopt an alternative parameterization and draw inferences about the equality of adjacent scores using the Bayesian approach. To achieve that, we have constructed a reversible jump Markov chain Monte Carlo algorithm for moving across models of different dimension and estimate accurately the posterior model probabilities which can be used either for model comparison or for model averaging. The proposed methodology is evaluated through a simulation study and illustrated using actual datasets.     

New aspects of the probabilistic evaluation of hypotheses and experience

The probabilistic corroboration of two or more hypotheses or series of observations may be performed additively or multiplicatively . For additive corroboration (e.g. by Laplace's rule of succession), stochastic independence is needed. Inferences, based on overwhelming numbers of observations without unexplained counterinstances permit hyperinduction , whereby extremely high probabilities, bordering on certainty for all practical purposes may be achieved. For multiplicative corroboration, the error probabilities (1 - Pr) of two (or more) hypotheses are multiplied. The probabilities, obtained by reconverting the product, are valid for both of the hypotheses and indicate the gain by corroboration.. This method is mathematically correct, no probabilities > 1 can result (as in some conventional methods) and high probabilities with fewer observations may be obtained, however, semantical independence is a prerequisite. The combined method consists of (1) the additive computation of the error probabilities (1 - Pr) of two or more single hypotheses, whereby arbitrariness is avoided or at least reduced and (2) the multiplicative procedure . The high reliability of Empirical Counterfactual Statements is explained by the possibility of multiplicative corroboration of “all-no” statements due to their strict semantical independence.