Influential observations in principal factor analysis

We propose a method for detecting influential observations in iterative principal factor analysis. For this purpose we derive the influence functionsI(x; LL ^T ) andI(x; ) for the common variance matrixT =LL ^T and the unique variance matrix , respectively, in the common factor decomposition =LL ^T + . A numerical example is given for illustration.The authors are grateful to Tomoyuki Tarumi and Atsuhiro Hayashi for their kind permission to use their software Seto/B for drawing Figures 1 and 2 and to anonymous reviewers for comments on the paper.     

Using item-specific instructional information in achievement modeling

The problem of detecting instructional sensitivity (item basis) in test items is considered. An illustration is given which shows that for tests with many biased items, traditional item bias detection schemes give a very poor assessment of bias. A new method is proposed instead. This method extends item response theory (IRT) by including item-specific auxiliary measurement information related to opportunity-to-learn. Item-specific variation in measurement relations across students with varying opportunity-to-learn is allowed for.This paper was presented at the 1987 AERA meeting in Washington, DC. This research was supported by grant OERI-G-86-003 from the Office of Educational Research and Improvement, Department of Education. The author thanks Michael Hollis and Chih-fen Kao for valuable research assistance, and appreciates valuable comments made by an anonymous reviewer.     

Algorithmic shaping and misbehavior in the acquisition of token deposit by rats

In two experiments, rats were trained to deposit ball bearings down a hole in the floor, using an algorithmic version of shaping. The experimenter coded responses expected to be precursors of the target response, ball bearing deposit; a computer program reinforced these responses, or not, according to an algorithm that mimicked the processes thought to occur in conventional shaping. In the first experiment, 8 of 10 rats were successfully shaped; in the second, 5 of 5 were successfully shaped, and the median number of sessions required was the same as for a control group trained using conventional shaping. In both experiments, “misbehavior,” that is, excessive handling and chewing of the ball bearings, was observed, and when the algorithmic shaping procedure was used, misbehavior could be shown to occur in spite of reduced reinforcement for the responses involved.     

A NEUROBIOLOGICAL ALTERNATIVE TO THE PERCEPTUAL REINFORCEMENT HYPOTHESIS OF STEREOTYPED BEHAVIOR: A COMMENTARY ON “SELF-STIMULATORY BEHAVIOR AND PERCEPTUAL REINFORCEMENT”

A perceptual reinforcement theory of stereotyped movements is advanced by Lovaas, Newsom, and Hickman (1987) in an effort to integrate a number of diverse observations about the origins and maintenance of this behavior. We, in turn, argue that the theory, as presented, is logically flawed and fails to take into account important biological findings and theory concerning pathological stereotyped acts. An alternative theory, derived primarily from neurological concepts, is briefly described.     

Unprogrammed effects of training high-status peers to interact with severely handicapped children.

We examined the effects of a peer initiation intervention with high- and low-status nonhandicapped students on the behavior of untrained peers toward handicapped students. In the context of a counterbalanced withdrawal design, high- and low-status nonhandicapped students were taught to direct social initiations to eight severely handicapped students during recess activities. The interactions of the high-status students resulted in higher levels of initiations by untrained peers toward the handicapped students than did the interactions of the low-status students. Social response levels were also differentially affected by the status of the peer initiator.     

A nonparametric approach for assessing latent trait unidimensionality

Assuming a nonparametric family of item response theory models, a theory-based procedure for testing the hypothesis of unidimensionality of the latent space is proposed. The asymptotic distribution of the test statistic is derived assuming unidimensionality, thereby establishing an asymptotically valid statistical test of the unidimensionality of the latent trait. Based upon a new notion of dimensionality, the test is shown to have asymptotic power 1. A 6300 trial Monte Carlo study using published item parameter estimates of widely used standardized tests indicates conservative adherence to the nominal level of significance and statistical power averaging 81 out of 100 rejections for examinee sample sizes and psychological test lengths often incurred in practice.The referees' comments were remarkably detailed and greatly enhanced the writeup and sensitized the author to certain pertinent issues. Discussions with Fritz Drasgow, Lloyd Humphreys, Dennis Jennings, Brian Junker, Robert Linn, Ratna Nandakumar, and Robin Shealy were also very useful.This research was supported by the Office of Naval Research under grant N00014-84-K-0186; NR 150-533, and by the National Science Foundation under grant DMS 85-03321.     

Reliability of test scores in nonparametric item response theory

Three methods for estimating reliability are studied within the context of nonparametric item response theory. Two were proposed originally by Mokken (1971) and a third is developed in this paper. Using a Monte Carlo strategy, these three estimation methods are compared with four classical lower bounds to reliability. Finally, recommendations are given concerning the use of these estimation methods.The authors are grateful for constructive comments from the reviewers and from Charles Lewis.     

Behavioral contrast in fixed-interval components: effects of extinction-component duration.

Seven albino rats were exposed to a multiple schedule of reinforcement in which the two components (fixed interval and extinction) alternated such that a presentation of the extinction component followed each fixed-interval reinforcement. In baseline sessions, the duration of the extinction component was constant and always one-third of the fixed-interval value. Probe sessions contained a probe segment in which the duration of the extinction component was increased; the response rate in fixed-interval components during the probe segment was compared with the response rate in the segments preceding and following the probe. The effect of increasing the duration of the extinction component was studied under three values of fixed interval: 30 s, 120 s, and 18 s, in three successive conditions. Response rate within fixed intervals was a direct function of duration of the extinction component. Pausing at the beginning of the fixed interval decreased as extinction duration increased. These effects were larger and more consistent for the shorter fixed-interval values (18 s and 30 s). These results indicate a functional relation between relative component duration and responding. For the component providing more frequent reinforcement, this could be stated as an inverse relationship between relative component duration and response rate. This relation is similar to findings regarding the ratio of trial and intertrial duration in Pavlovian conditioning procedures, and suggests that behavioral contrast may be related to Pavlovian contingencies underlying the multiple schedule.     

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.