Search

The paper presents results derived from a series of simulations of signed networks (i.e., networks containing positive and negative links) subject to a balancing process. The dynamics are followed in a space defined by variations in the proportion of positive links introduced and the tolerance to imbalance. The authors claim to have discovered boundaries in the space separating it into three distinct regions. One region probably displays the characteristics of self-organized criticality, which may have implications for the theory of extreme events. [ABSTRACT FROM AUTHOR]

This exploratory paper proposes a probabilistic approach to investigating college student attrition. Through the use of attributable risk, the top risk factors for attrition are identified and ranked. Then the issue of confounding risk between two risk factors is addressed by the means of relative attributable risk. Though it is only illustrated by the eight-year attrition data collected at a private institution, the proposed technique is comparatively simple for institutional researchers to test on their own data. [ABSTRACT FROM AUTHOR]

This paper considers the statistical inference of structural properties in organizations. Specifically, the sampling probability distributions of the number of links and the cumulative distribution of the number of components are determined. Probability estimates of both the number of links and the number of components are then found. [ABSTRACT FROM AUTHOR]

This paper compares three methods (sequel, cardinal, maximal) for constructing a weak order from a partial order on a finite set. The constructed weak orders include the partial order. To evaluate the methods, several different selection disciplines were used to stochastically generate partial orders from a fixed linear order. The error of a weak order which includes a generated partial order is a function of the number of ordered pairs added to the partial order to get the weak order which are the reverse of ordered pairs in the fixed linear order. In all cases, the sequel and cardinal mean errors were much lower than the maximal mean error. In most but not all cases, the cardinal mean error was lower than the sequel mean error. [ABSTRACT FROM AUTHOR]