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.     

Examining differential item functioning due to item difficulty and alternative attractiveness

A method for analyzing test item responses is proposed to examine differential item functioning (DIF) in multiple-choice items through a combination of the usual notion of DIF, for correct/incorrect responses and information about DIF contained in each of the alternatives. The proposed method uses incomplete latent class models to examine whether DIF is caused by the attractiveness of the alternatives, difficulty of the item, or both. DIF with respect to either known or unknown subgroups can be tested by a likelihood ratio test that is asymptotically distributed as a chi-square random variable.     

Provision of step-level public goods with continuous contribution

Twelve groups of five subjects each participated in a nonco-operative game in which each member of a group receives the same endowment and must then decide independently and anonymously how much of it to contribute to the group benefit. Regardless of the size of his or her contribution, each member receives the same reward if, and only if, the sum of contributions is equal to or larger than a prespecified provision threshold. The results show that the level of contribution depends on the provision threshold, and that it increases when contributions are not restricted to be all-or-none. We present, discuss, and competitively test two models for this class of social dilemmas, one postulating maximization of expected utility and the other yielding an equitable solution.     

Incorporating prior theory in covariance structure analysis: A bayesian approach

A Bayesian approach to the testing of competing covariance structures is developed. The method provides approximate posterior probablities for each model under consideration without prior specification of individual parameter distributions. The method is based on ayesian updating using cross-validated pseudo-likelihoods. Given that the observed variables are the samefor all competing models, the approximate posterior probabilities may be obtained easily from the chi square values and other known constants, using only a hand calculator. The approach is illustrated using and example which illustrates how the prior probabilities can alter the results concerning which model specification is preferred.     

GENERALIZED LANGUAGE LEARNING BY CHILDREN WITH SEVERE MENTAL RETARDATION: EFFECTS OF PEERS' EXPRESSIVE MODELING

In this study, we investigated the conditions that contribute to observational learning of generalized language in children with severe mental retardation. Matrix-training strategies were used to teach 6 children with mental retardation to combine known words into two- or three-word utterances consistent with syntactic rules. Subsequently, the children learned two or more unknown words concurrently, inducing word-referent relations consistent with these word order rules. Generalized learning of responses not taught directly was shown to be under experimental control using a multiple baseline design across submatrices. Expressive modeling of only four or five responses was sufficient to promote recombinative generalization in the expressive and receptive modalities. Thus, 95% to 98% of subjects' learning was attributed to generalization processes. This study demonstrates how the efficiency of language training with children with mental retardation might be enhanced by coupling observational learning and matrix-training strategies.     

Temporal proximity and reinforcement sensitivity in multiple schedules

Distributions of reinforcers between two components of multiple variable-interval schedules were varied over a number of conditions. Sensitivity to reinforcement, measured by the exponent of the power function relating ratios of responses in the two components to ratios of reinforcers obtained in the components, did not differ between conditions with 15-s or 60-s component durations. The failure to demonstrate the “short-component effect,” where sensitivity is high for short components, was consistent with reanalysis of previous data. With 60-s components, sensitivity to reinforcement decreased systematically with time since component alternation, and was higher in the first 15-s subinterval of the 60-s component than for the component whose total duration was 15 s. Varying component duration and sampling behavior at different times since component transition may not be equivalent ways of examining the effects of average temporal distance between components.     

SERIOUS DELINQUENT BEHAVIOR AS PART OF A SIGNIFICANTLY HANDICAPPING CONDITION: CURES AND SUPPORTIVE ENVIRONMENTS

The past 20 years have been productive ones for the field of applied behavior analysis. A brief review of our own efforts during this period reveals that we have accomplished several but not all of our goals for the Teaching-Family approach. In this context, we note that the setting of realistic and appropriate goals is important for the field and for society. Moreover, we suggest that the realistic goal for some persons with serious delinquent behavior may be extended supportive and socializing treatment rather than permanent cure from conventional short-term treatment programs. We base this suggestion on the accumulating evidence that serious delinquent behavior may often be part of a significantly disabling and durable condition that consists of multiple antisocial and dysfunctional behaviors, often runs in families, and robustly eludes effective short-term treatment. Like other significant disabilities such as retardation, autism, and blindness, the effects of this condition may be a function of an interaction of environmental and constitutional variables. We argue that our field has the wherewithal to construct effective and humane long-term supportive environments for seriously delinquent youths. In this regard, we explore the dimensions, rationales, logistics, and beginnings of a new treatment direction that involves long-term supportive family treatment. We contend that such supportive families may be able to provide long, perhaps even lifetime, socializing influences through models, values, and contingencies that seem essential for developing and maintaining prosocial behavior in these high-risk youths.     

CORRESPONDENCE TRAINING: AN EXAMPLE OF RULE-GOVERNED BEHAVIOR?

Two groups of six mentally retarded adults were exposed to either a “say/do” correspondence training program or a “do only” reinforcement procedure to evaluate the suggestion of Rogers-Warren and Baer (1976) that reinforcement of the nonverbal target behavior in the absence of the relevant verbal behavior may account for the behavior changes seen in correspondence training. The participants worked in an experimental setting on a variety of manipulatory responses leading to various auditory and visual consequences. Analysis of individual patterns of responding indicated no apparent differences between the groups during training; four individuals in each group appeared to develop generalized correspondence skills. We conclude that the outcome of correspondence training may not necessarily be verbal regulation of behavior as is assumed. Rather, we suggest that the notion of rule-governed behavior can best account for the type of behavior changes seen in correspondence studies.     

Latent trait models and dichotomization of graded responses

This paper discusses thecompatibility of the polychotomous Rasch model with dichotomization of the response continuum. It is argued that in the case of graded responses, the response categories presented to the subject are essentially an arbitrary polychotomization of the response continuum, ranging for example from total rejection or disagreement to total acceptance or agreement of an item or statement. Because of this arbitrariness, the measurement outcome should be independent of the specific polychotomization applied, for example, presenting a specific multicategory response format should not affect the measurement outcome. When such is the case, the original polychotomous model is called compatible with dichotomization.A distinction is made between polychotomization or dichotomization before the fact, that is, in constructing the response format, and polycho- or dichotomization after the fact, for example in dichotomizing existing graded response data.It is shown that, at least in case of dichotomization after-the-fact, the polychotomous Rasch model is not compatible with dichotomization, unless a rather special condition of the model parameters is met. Insofar as it may be argued that dichotomization before the fact is not essentially different from dichotomization after the fact, the value of the unidimensional polychotomous Rasch model is consequently questionable. The impact of our conclusion on related models is also discussed.