Existing approaches to combining multiple forecasts generally offer either theoretical richness or empirical robustness, but not both together. In this paper, we propose a new method for combining forecasts which attempts to overcome this imbalance. The method allows easy inclusion of relevant subjective and empirical information about the forecasts, while providing weights which are: (i) intuitively meaningful, and (ii) not dependent upon large numbers of observations of prior forecast accuracy. Results of a simulation experiment show the method to be highly robust, and significantly superior to existing approaches under many conditions.

Consider a job shop that is modelled as an open queueing network of the Jackson (Jackson, J. R. 1957. Networks of waiting lines. Oper. Res. 5 518–521.) type. All work stations in the shop have the same number of parallel servers. Two problems are studied: the loading of stations and the assignment of servers, which are represented by loading and assignment vectors, respectively. Majorization and arrangement orderings are established to order, respectively, the loading and the assignment vectors. It is shown that reducing the loading vector under majorization or increasing the assignment vector under arrangement ordering will reduce the congestion in the shop in terms of reducing the total number of jobs (in the sense of likelihood ratio ordering), the maximum queue length (in the sense of stochastic ordering), and the queue-length vector (in the sense of stochastic majorization). The results can be used to support production planning in certain job shops, and to aid the design of storage capacity.

A crucial undertaking in research on organizations is to obtain valid estimates of the organization-level constructs and their relationships to one another. One organizational research methodology has multiple individuals within each firm act as informants and report their perceptions of these constructs. Drawing upon past work, a comprehensive, confirmatory measurement approach is presented that enables delineation of the organizational constructs of interest from potential sources of measurement error in multiple informant reports. A confirmatory composition model first specifies the relationship between measurements taken from multiple informants and organization-level indicators of organizational constructs. This model provides an explicit representation of the informants' perceptual agreement with respect to each measure, the bias in their responses due to their particular perspectives on the firm, and random measurement error. Measures which demonstrate significant perceptual agreement across informants are retained as organization-level indicators. A confirmatory measurement model is then presented that relates the defined indicators to their posited underlying organizational constructs, with the constructs allowed to freely inter-correlate. Finally, a structural model relates the constructs to one another as specified by some theory. The structure model in conjunction with the measurement model permits a comprehensive assessment of the construct validity of the organizational properties. An extension of the approach is also discussed where the assessment of commonalities and differences between different kinds of organizations with respect to some theoretical model is of interest. An illustration of the confirmatory approach is given, drawing upon data from a study of distributor firm and manufacturer firm working relationships.

Biases that reflect the economic worth of uncertain contingent claims occur in many financial markets. Parimutuel betting at racetracks is one such market with ample data to investigate such biases. The total wagering market is about $10 billion per year in North America. The configuration of racetracks leads to an advantage for horses breaking from post positions near the rail, especially for tracks with small circumferences. Can the bettor make profits with knowledge of this bias? To investigate, we utilize data from 3,345 races involving over $300 million in wagers from 1982, 1983 and 1984 on win and exotic bets at Exhibition Park in Vancouver where this bias should be strong. The results indicate that the bias exists but the prices adjust to fully negate the potential gains from the bias.

This article is concerned with counting and identifying those customers who are still active. The issue is important in at least three settings: monitoring the size and growth rate of a firm's ongoing customer base, evaluating a new product's success based on the pattern of trial and repeat purchases, and targeting a subgroup of customers for advertising and promotions. We develop a model based on the number and timing of the customers' previous transactions. This approach allows computation of the probability that any particular customer is still active. Several numerical examples are used to illustrate applications of the model.

This note corrects an error in van Nunen and Puterman (van Nunen, J. A. E. E., M. L. Puterman. 1983. Computing optimal control limits for GI/M/s queueing systems with controlled arrivals. Management Sci. 29 725–734.) and extends some of their results to the average reward case.

We develop a method for obtaining approximate steady-state probabilities for large multi-echelon reparable item inventory systems modeled as non-Jacksonian Markovian networks with finite state space. The approximation involves decomposing the network model into smaller overlapping local subnetwork models, solving them in “isolation” and iterating back and forth among the subnetwork models until convergence is obtained. Numerical results show that the method is quite accurate and efficient for this application.