Although decision-making represents a fundamental issue in management information systems (MIS), obtaining accurate assessments of the factors affecting employees' decisions may be difficult using traditional methods such as ratings and rankings. Policy capturing, a little-used method in MIS, represents a potentially important alternative to more traditional methods. After demonstrating that policy capturing has been underutilized in MIS, the paper illustrates the use of policy capturing in two decision-making contexts—computer training and software selection. These two studies contrast policy capturing results with more traditional methods, and draw implications for research.

Coding capacity is considered for a class of additive dimension-limited channels. The channels may be with or without memory, stationary or nonstationary. The constraint is partially given in terms of an increasing family of finite-dimensional subspaces. A general expression for the capacity is obtained, which depends on the family of subspaces and the relation between the noise covariance and the covariance giving the energy-frequency constraint on the transmitted signal. This result holds for all classical discrete-time Gaussian channels and for continuous-time Gaussian channels with fixed time of transmission, so long as a peak energy constraint is used on the codewords. The expression also provides upper bounds for a class of non-Gaussian channels. Several results are obtained that aid in calculation of capacity for specific applications. For this class of channels, it is shown that coding capacity is equal to information capacity. Error bounds are given for Gaussian channels.>

The averaged information capacity of the mismatched stationary continuous-time Gaussian channel is considered, in the limit as the observation time becomes infinite. It is required only that the signal satisfy a reproducing kernel Hilbert space constraint on expected energy. This requires the, signal energy to be distributed in regions where the noise energy, is not too small, and is the weakest condition that provides finite capacity. In the case when both the noise covariance and the filter on the message are expressed by rational functions, the results complement those previously obtained by Gallager and Holsinger (1968), and the combination gives a complete solution to the capacity problem. The treatment provides a desirable generality on the class of transmitted signals that is not present in previous treatments.>

Information capacity is determined for the additive Gaussian channel when the constraint is given in terms of a covariance different from that of the channel noise. These results, combined with previous results on capacity when the constraint covariance is the same as the noise covariance, provide a complete and general solution for the information capacity of the Gaussian channel without feedback. The results are valid for both continuous-time and discrete-time channels and require only two assumptions: the noise is due to a stochastic process with sample paths having finite energy over the observation period (w.p.1), and the constraint is given in terms of a Hilbert space norm. Such a constraint is implicit in any constraint giving finite capacity.

Bounds on the coding capacity of Gaussian channels are obtained when the power constraint on the signal is mismatched to the channel noise. In the case of some feedback channels in which the noise has a Cramér-Hida representation of finite multiplicity, an exact expression for the coding capacity is given.

The hypothesis of multivariate normality is frequently made for sonar data. Signal processing systems where design is based on this assumption may perform poorly if the assumption is not satisfied. Tests for multivariate normality are thus of interest. A test is described that can be used when the covariance matrix and mean vector of the data are known.

The "Bryn Processor" is a well-known array processor; its design depends on spectral densities of the signal and noise processes. Array output signal-to-noise ratio can be decreased because of errors in sampling times: (1) When array design parameters are estimated from sampled data; (2) when operating on sampled input data. We model errors in sampling times as a discrete-parameter random process. Expressions for array output SNR of the Bryn processor are determined for the two situations listed above. Numerical results are presented for some specific assumptions on signal, noise, and sampling error.

The deflection of a bounded quadratic-linear test statistic is considered for the following binary detection problem. HypothesisH_{1}--received waveform is a sample function from a random process with known covariance and mean functions, but unknown probability distributions, versusH_{2}--received waveform is a sample function from a Gaussian process (noise) having known covariance and mean functions. Sample functions are assumed to belong to a real and separable Hilbert space. The test statistic is assumed to be the sum of a bounded quadratic operation and a bounded linear operation on the data. Necessary and sufficient conditions for the deflection to be bounded over all non-null bounded quadratic-linear operations are given, and additional results are obtained under the assumption that the deflection is bounded. Several relations are shown to exist between the deflection problem and the optimum discrimination problem when both processes are Gaussian. In particular, it is shown that nonsingular discrimination occurs if and only if a generalized deflection is bounded, and that in some cases the problem of realizing the log-likelihood ratio is equivalent to the problem of attaining the least upper bound for the deflection.