A psychological profile of juvenile firesetters in residential treatment

Psychological test data of 30 known juvenile and adolescent firesetters were compared with those of a matched control group, within a residential treatment center, to distinguish the identifying characteristics of each group. Hypotheses based on 80 variables found in the literature on firesetting were treated by using a standard test equivalent for each variable. Of the hypotheses, 14 were found to differentiate significantly the two groups, but not all in the predicted direction. Four characteristics were found to be sufficient to discriminate between the groups with greater efficiency than any other combination of variables. A predictor equation is set forth.     

Monitoring antitobacco sentiment among community leaders

Two public health objectives, one methodological and the other substantive, are realized in the present study. First, average survey ratings are replaced by negated mean cumulative logits (MCLs), which have the advantage of interitem commensuration. Second, these negated MCLs improve the analysis of two Florida tobacco control surveys of community leaders. Because of their common logit metric, negated MCLs for different item types are compared on their expressed antitobacco sentiment. They are also averaged to construct a precise indicator of overall sentiment. These results show that in 1998 and 1999, community leaders were most averse to selling to minors, kids buying cigarettes, and restaurant and workplace smoking. The latter aversion anticipated voters' overwhelming support of the 2002 Florida constitutional amendment prohibiting smoking in restaurants and workplaces. The authors hope that this analysis for antitobacco programs will improve other substance-abuse monitoring with surveys whose items are yet unwritten.     

Partitioned algorithms for maximum likelihood and other non-linear estimation

There are a variety of methods in the literature which seek to make iterative estimation algorithms more manageable by breaking the iterations into a greater number of simpler or faster steps. Those algorithms which deal at each step with a proper subset of the parameters are called in this paper partitioned algorithms. Partitioned algorithms in effect replace the original estimation problem with a series of problems of lower dimension. The purpose of the paper is to characterize some of the circumstances under which this process of dimension reduction leads to significant benefits.Four types of partitioned algorithms are distinguished: reduced objective function methods, nested (partial Gauss-Seidel) iterations, zigzag (full Gauss-Seidel) iterations, and leapfrog (non-simultaneous) iterations. Emphasis is given to Newton-type methods using analytic derivatives, but a nested EM algorithm is also given. Nested Newton methods are shown to be equivalent to applying to same Newton method to the reduced objective function, and are applied to separable regression and generalized linear models. Nesting is shown generally to improve the convergence of Newton-type methods, both by improving the quadratic approximation to the log-likelihood and by improving the accuracy with which the observed information matrix can be approximated. Nesting is recommended whenever a subset of parameters is relatively easily estimated. The zigzag method is shown to produce a stable but generally slow iteration; it is fast and recommended when the parameter subsets have approximately uncorrelated estimates. The leapfrog iteration has less guaranteed properties in general, but is similar to nesting and zigzagging when the parameter subsets are orthogonal.     

Predicting outcomes: Sports and stocks

Many gamblers and most fans, players, and coaches offer causal explanations for long runs of good or bad performance in sports and financial analysts are quick to offer explanations for the daily performance of the stock market. The records of professional basketball and baseball teams and the Dow Jones daily closing average for a ten year period were evaluated for trends (streaks). The records of teams were also evaluated to assess whether the record against opponents, the home court or home field advantage, and—for baseball teams—the record of the winning and losing pitcher (excluding the current game) predicted the outcome of individual games. Recent performance is, at best, a very weak predictor of current performance and the three best predictors for baseball (pitching, home field, and record against opponent) together accounted for only 1.7% of the variance in the outcomes of individual games. We overestimate our ability to predict. This overconfidence is likely to play a role in maintaining gambling behaviors.     

Measuring subjective social indicators

A logistic response model is proposed for measuring social indicators based upon survey rating scales. A close fit of the model to national life-quality data is demonstrated. Moreover, the model is constrained so as to provide an explicit formulation for item averaging with respect to group aggregates. This item redundancy provides measures for social constructs in a manner analogous to traditional procedures at the individual level of measurement. Further applications of this principle to opinion and mental health monitoring are discussed.     

Estimating operational costs in an electric utility

Generation expansion planning in the electric utility industry requires consideration of uncertainties in both the demand and supply of electric power. The expected demand is usually expressed via a load-duration curve, while, on the supply side, each generating unit has a given nameplate capacity and a predicted reliability. This paper focuses on considerations of the supply-side uncertainties and their effects on estimating operating costs in electric utility planning. However, the methods and analysis developed in this paper may be applicable to a wider class of production planning problems which deal with any nonstorable product with time varying demand. Two methods for estimating the energy generation from each generating unit are compared. The first is the method of probabilistic simulation, while the second involves a heuristic technique usually denoted the derating method. A bias inherent in the derating method is examined by comparing it with a probabilistic simulation method. The bias is examined for various load curve shapes. In certain cases, a closed form expression for the bias is obtained. However, a closed form expression of the bias for an arbitrary load curve is difficult to achieve. In these situations some examples are studied in which the trend of the relative bias among plants in the loading order is examined. Finally, the bias is examined using actual 1977 load and supply data for some New England utilities.     

Territorial invasion and aggression in young children

Forty black first- and second-grade children were given instructions intended to induce possessiveness for both a designated toy and a play area, along with an opportunity to play with the toy in the area. Each subject then surreptitiously observed an experimental confederate dressed in a clown costume play with either the designated toy or another toy, while either within or outside the child's play area. While observing the invasion, the child had 10 opportunities (trials) to deliver shock to the confederate. Girls who observed the confederate playing with the designated toy delivered more shocks and waited fewer trials to deliver the first shock than did the other girls. Similar effects were produced by the confederate's use of the subject's play area. The manipulations produced no significant effects upon the boys' aggression.