Mixtures of conditional mean- and covariance-structure models

Models and parameters of finite mixtures of multivariate normal densities conditional on regressor variables are specified and estimated. We consider mixtures of multivariate normals where the expected value for each component depends on possibly nonnormal regressor variables. The expected values and covariance matrices of the mixture components are parameterized using conditional mean- and covariance-structures. We discuss the construction of the likelihood function, estimation of the mixture model with regressors using three different EM algorithms, estimation of the asymptotic covariance matrix of parameters and testing for the number of mixture components. In addition to simulation studies, data on food preferences are analyzed.The authors are grateful to Donald B. Rubin and Michael E. Sobel for critical reading of a first draft of this paper and to three anonymous reviewers ofPsychometrika for their helpful comments and suggestions. The research of the first and the third author was supported by a grant from the Deutsche Forschungsgemeinschaft.     

Remembering children's emotions: sources of concordant and discordant accounts between parents and children.

Parents were asked to recall recent events that had evoked happiness, sadness, anger, and fear in their children. Children (N = 77, 2.3-6.6 years) indicated whether they remembered each event, and if so, they described the event and how it had made them feel. Agreement between parent and child concerning how the child felt varied as a function of emotion. Children agreed with their parents' emotion attributions most often for events that parents recalled as having evoked happiness and sadness, less often for fear, and least often for anger. Children disagreed with parents' attributions of happiness and sadness most often when parents and children differed concerning the attribution of children's goals. Discordant reports about children's anger were most frequent when parents and children reported conflicting goals. Discordant reports about fear were most frequent when parents and children focused on different parts of the temporal sequence surrounding the event.     

A scale for the psychological magnitude of number

A scale of the “psychological magnitude” of number was constructed from similarity ratings of the 45 number pairs that can be obtained from a set of 10 integers. A nonmetric analysis of these similarity ratings showed that “psychological number” was a power function of number.     

Writing about psychoanalysis: II. Analysts who write, patients who read

When we make a decision to lecture, or to write for publication, we have begun an activity that will affect every one of our patients. If they attend our lectures, or read our published works, they cannot help being influenced. Even if they do not, they will have some reaction to the focusing of our interest on particular phases of the analysis. This will be true whether or not we intend to present clinical material. These considerations inevitably affect the analyst himself, sometimes inhibiting his writing altogether, sometimes constricting its scope, or worst of all, degrading its quality. The analyst's writing is also almost certain to influence whether and with what success his patient may choose to write. Lest these considerations suggest that we avoid writing altogether, I emphasize that such a sacrifice is by no means desirable or necessary. Problems of discretion will arise, and may in some cases be virtually insoluble; but there are ways of writing that can overcome most such objections. The analyst's interest in writing should be recognized and its effects, if any, brought into the analysis; so long as the writing itself is done with due regard to the conduct of the analysis and the welfare of the patient, it need have no adverse effects on either.     

Sexual assault history and use of health and mental health services

A history of sexual assault may be associated with increased current use of mental health and medical services because of the psychologically and physically disruptive consequences of assault. To test this hypothesis, we estimated rates of mental health and medical services use among 2560 randomly selected community residents, 343 of whom had been sexually assaulted. Sexual assault was associated with seeking both forms of care. Controls for demographic variables, psychiatric diagnosis, health status, and insurance suggested that assault increases use indirectly, through poor mental and physical health. Uninsured, assaulted respondents were especially likely to consult medical providers. Respondents assaulted during childhood were particularly likely to seek mental health care. Assault was more common among mental health service users than nonusers, and among women using medical services compared to female nonpatients. The high prevalence of assault among service users underscores the need for providers to recognize and treat sexual assault-related problems.     

Constraints on effective pictorial and verbal elaboration

Three experiments investigated the conditions under which pictures facilitate learning. In Experiment 1, confusing verbal relationships were supplemented with pictures that illustrated the key concepts in each verbal relationship (base pictures), illustrated the key concepts in more elaborate arbitrary relationships (pictures of arbitrary elaborations), or illustrated the key concepts in ways that helped clarify the verbal relationships (pictures of explanatory elaborations). All three types of pictures facilitated the retention of the verbal relationships, although pictures of explanatory elaborations were superior to other types of visual illustrations. In Experiment 2, the facilitative effects of base pictures depended on a schematically unique illustration of the key concepts in a single image. In Experiment 3, picture facilitation effects were constrained by the type of verbal elaborations that accompanied the pictures. Specifically, base pictures and pictures of arbitrary elaborations did not facilitate the retention of confusing verbal relationships that were elaborated with explanatory information, and actually interfered with the retention of those explanatory elaborations. The implications of these results are discussed.     

Constraints on spontaneous transfer in problem-solving tasks

In three experiments, we investigated the conditions under which relevant knowledge is spontaneously transferred to problem-solving tasks. Subjects were presented with key concepts that could help them solve problems presented at a later time in the experiment. The key concepts were embedded in statements that had surface structures similar or dissimilar to those of the problems and that emphasized relevant or irrelevant properties of the key concepts for the problemsolving task (contextual relevance). The results indicated that the spontaneous transfer of clue information to subsequent problem-solving tasks is affected by the contextual relevance and the surface-structure similarity of the clue statements to the problems. The findings also suggested that the surface-structure similarity and the contextual relevance of clue statements differentially affect the accessibility and appropriate application of key concepts in problem-solving tasks. The implications of these results for understanding the transfer of knowledge in problem-solving tasks are discussed.     

MMPI predictors of mania among psychiatric inpatients

MMPI data from 64 patients with a diagnosis of manic-depressive illness, manic type, were compared with MMPI data from patients in two comparison groups--64 patients with a psychotic diagnosis other than manic-depressive illness, and 64 patients with a variety of psychiatric diagnoses. Manic patients had higher Ma scale scores for MMPI scales that assess personal distress and interpersonal difficulties (e.g., D and Si). Discriminant analysis, with the Ma, D, and Si scales as predictors, correctly classified as manic or not manic 82.5% of the patients in the derivation sample and 74.2% of the patients in the cross-validation sample. Two high-point pairs, Sc-Ma/Ma-Sc and Pa-Ma/Ma-Pa, occurred in the MMPI profiles of almost half of the manic patients but were rarely found among the profiles of other patients. The results of this study support the use of the MMPI in identifying manic patients, particularly when discriminating between mania and other types of psychosis.