A Simple Simulation Technique for Nonnormal Data with Prespecified Skewness,Kurtosis, and Covariance Matrix

We present and investigate a simple way to generate nonnormal data using linear combinations of independent generator (IG) variables. The simulated data have prespecified univariate skewness and kurtosis and a given covariance matrix. In contrast to the widely used Vale-Maurelli (VM) transform, the obtained data are shown to have a non-Gaussian copula. We analytically obtain asymptotic robustness conditions for the IG distribution. We show empirically that popular test statistics in covariance analysis tend to reject true models more often under the IG transform than under the VM transform. This implies that overly optimistic evaluations of estimators and fit statistics in covariance structure analysis may be tempered by including the IG transform for nonnormal data generation. We provide an implementation of the IG transform in the R environment.     

An unobtrusive measurement method of the horizontal gaze angle

A horizontal gaze angle measurement device is introduced. By combining a photoelectric viewing device to measure the horizontal eye angle with a similar head angle measurement device, it is possible to measure the horizontal gaze angle without using a headrest. After discussion of circuit diagrams and measurement principles, it is shown that the measurements made with the device yield a reasonable precision. The mean absolute measurement error is below 1°. This inexpensive and unobtrusive device covers a visual field of about 20° and can be used in parallel with many tasks. Further, data on the successful application of the device in a driving simulation setting are discussed.     

Expectancy‐effects without expectancies: Illusory correlations based on cue‐overlap

Expectancy‐based illusory correlations have been observed in numerous experiments. Simulations using the connectionist computer model BIAS (Fiedler, 1996 ) show that such illusory correlations may not always reflect expectancy biases but influences of similarity via cue‐overlap. Cue‐overlap means that some of the probabilistic cues that indicate the presence of one variable are also indicative of another variable. In an experiment, participants learned two novel concepts pertaining to a fictitious painter and a fictitious artistic style in separate runs. Both concepts were defined by multiple probabilistic cues observable in paintings. For half the participants, the cue systems underlying the perception of the two concepts overlapped, whereas for the other half they did not. In addition, we manipulated whether or not participants expected a positive contingency between artist and artistic style. In the second part of the experiment, a series of paintings was presented that constituted an objective zero correlation between artist and artistic style. Participants' subsequent contingency judgments were assessed by direct and indirect measures. Data analyses revealed main effects for expectancy induction and cue‐overlap but no interaction on the direct measure and nearly identical results on the indirect measure. Thus, cue‐overlap and expectancy induction independently triggered the development of illusory correlations. Copyright © 2000 John Wiley & Sons, Ltd.