FLOOD09697 - Laserfiche WebLink

Home Browse Search

Table 5.3. Terms of agency equationsQ (Equations III, IV, and VI). (1) Presence of a Flood Control Problem (IFPROB) (2) Agency Concern to Include Low Cost in Flood Control Project (IACCOS) by Benefit-COst Ratio of Flood Con- trol Proposal (BECORA) (3) Agency Concern to Include Low Cost in Flood Con- trol Project (lACCOS) by One over the Cost of the Proposal (I +COSPRO) (4) Agency Concern to Include Effectiveness in Flood Control Proposal (IACEFF) by Average Annual Bene- fit of Flood Control Proposal in Dollars (A VEBEN) (5) Agency Concern to Include Pleasing Aesthetics in Flood Control Proposal (lACAES) by Original Judges Esti- mate of Aesthetics Effect of Flood Control Proposal (OJEAES) (6) Agency Concern to Include Recreation in Flood Con- trol Proposal (lACREC) by Original Judges Estimate of Recreational Effect of Flood Control PIOpOSal (OJERECl (7) Agency Concern to Include Least Detrimental Envir- onmental Effect in Flood Control Proposal (IACECO) by Original Judges Estimate of Ecological Effect of Flood Control Proposal (OJEECO) (8) Importance of Other Agency Opinion to Agency (AGEAGE) by Other Agency Opinion of Proposal (OTHEVE-3) (9) Importance of Public Opinion to Agency (PUBAGE) by Mean Public Evaluation of Proposal (PUBPRO-3) aTerms consist of one or more variables. cal of one of the variables in order to interact willing- ness to pay with a variable that increased for lower cost proposals. Since the range of potential values of COSPRO is very large, II was not feasible to subtract COSPRO from a specified number in order to reverse the direction of the values. Dividing one by the figure seemed the most reasonable method to obtain an in- dicator oflow cost. The relative sizes of the variables affects the unstandardized, but not the standardized, coefficients. The values of Ihe reciprocal of COSPRO are small since the values of COSPRO are large.12 This is the reason why the unstandardized coefficient of Term 3 is so much larger than any others in the agency equations. The effect of an interaction term upon the de- pendent is unaffected by the direction in which the variables are defined so long as both variables have the same Iype of relationship with the dependent vari- able, direcI or inverse. This is necessary so that when they are combined in one term, the effect of one does not tend to cancel but rather to reinforce the effect of the other variable. Most variables in acceptance functions were defined so that the relationship with the dependent variable would be direct, i.e., a higher value of the variable would mean greater acceptability 12T1le values of l/COSPRO are not as small as would be expected because the figures entered for COSPRO were costs in millions of dollars. of the proposal. Consequently, the signs of the inter- action terms would theoretically be positive. I 3 Anomalies such as Ihe negative signs of the first two lerms in Equation III (5.9) can occur for any of several reasons. Among the possible causes are omis- sion of important variables, poor measurement, inade- quate sample, and an incorrect form of the model. Asswning the relatlonships for these terms are correct, this leaves the first Ihree possibilities of which the first may be the most cogent in this case. One problem may be that the conceptual model of the decision process (Chapter IV) was not followed exactly in measuring of variables. As signified by the name. "Importance Factors, II the companion variables to the characteristics of proposals in Type III terms and to the evaluations by other groups in Type IV terms, should measure the importance of the compan- ion factors in evaluation of flood control proposals. As such, they would have no effect direct or indirect, on evaluation except in relation to the characteristics and evaluations. If a factor had a direcl relationship with the dependent variable, then the term would also. If a factor had an inverse relationship with the evaluation, regression analysis would produce a nega- tive coefficient automatically. There would be no need to predetermine the direction of the interacting variables.14 Assessing Ihe importance factors would have been easier than measuring the variables used. However. this assessment of importance factors was not done as directly as could have been. Attitudes were measured rather than the direct importance of a characteristic in flood control proposal evaluation and the amount of that characteristic present in the pro- posal being considered. Direct measurements would require a straight rating technique or ratio scaling. As it is, acceptance functions are interactions between various attitudes and expected impacts of a proposal. The coefficients when Equation III (5.9) is ap- plied to a plarUling agency would not have the sarne effects as when the equation is applied for the initial evaluation by the decision agency. This is because the values of the coefficients shown are multiplied by 13The desire to have all acceptance function positive re- sulted in difficulties both in the agency and population evaluation equations with the cost acceptance function. In the future, the companion variable should be measured such that COS- PRO can be entered without modification into the evaluation equations. See discussion of IF statement Cs. 14The reason for establishing importance factor values rather than letting the regression analysis specify regression co- efficients on the factors alone is in the hope of making the equa- tion more general. Importance factors vary from person to person and from group to group. Establishing an equation with set importance factors would preclude generality. The concept of acceptance functions accounts for the variability. See discussion of Type III terms in this chapter and decision process section of Chapter IV. 66