Predictive Validity of Central-Composite Design Regression Equations

Abstract

The predictive validity of the Mills and Williges (1973) empirically derived prediction equations of single operator performance in a simulated surveillance system was assessed by measuring 16 additional data points on the same four subjects participating in the original study. Correlations between predicted and observed performance on 16 points augmented to the design compared favorably with estimated shrunken multiple-correlation coefficients. In addition, the averages of each of the 16 additional treatment conditions were compared to the 95% confidence interval of the predicted values using the Mills and Williges (1973) regression equations. The 16 data points were also chosen such that a supplementary factorial analysis of variance could be conducted on the data. Comparisons were made between the analysis of variance and the multiple-regression analysis. It was concluded that the response surface methodology procedures for developing overall prediction equations of human performance demonstrate a high degree of predictive validity.

Get full access to this article

View all access options for this article.

References

Herzberg

P. A.

The parameters of cross-validation. Psychometric Monographs Supplement, 1969, 34, No. 16.

J. C. R.

Statistical inference II. The multiple regression and its ramifications. Ann Arbor: Edwards Brothers, 1964.

Lord

F. M.

Novick

M. R.

Statistical theories of mental test scores. Chapter 13. The selection of predictor variables. Reading, Mass.: Addison-Wesley, 1968, 284–301.

Mills

R. G.

Williges

R. C.

Performance prediction in a single-operator simulated surveillance system. Human Factors, 1973, 15, 337–348.

Williges

R. C.

North

R. A.

Prediction and cross-validation of video cartographic symbol location performance. Human Factors, 1973, 15, 321–336.