John S. Donovan
Published: 1976
Total Pages: 100
Get eBook
This thesis analyzes the quantitative and qualitative criteria of workoad factors that are used in determining manpower requirements. Three qualitative characteristics were recommended. First, the workload factor must have a causal relationship with the manpower requirement. Second, the factor should be predictable, i.e., it should be a good predictor of manpower and future values of the factor should be easily and economically obtained. And, finally, dummy variables should be used so that more causal factors can be employed in the model. Four quantitative criteria were identified. First, the standard error of the estimate as a percent of the dependent mean is the best estimate as to the precision of the model when used in conjunction with the coefficient of determination. Second, multivariate analysis will reduce stochastic error and increase the predictive properties of the model. Third, when the number of variables precludes examination of all possible regressions, a stepwise regression program with low acceptance limits should be used. And finally, in analyzing the effects of high collinearity, varables should not be eliminated from the model for failure of the 't' -test, if the variable is felt to have a causal relationship and it reduces the error sum of the squares. (Author).