FLOOD09697 - Laserfiche WebLink

Home Browse Search

mined by considering factors individually and collect- ively (Narayana et al., 1970, Namboodiri et aI., 1975). Single variable relationships are readily established with limited data. Multivariable relationships are more complicated, and the complexity increases with the number of variables. When the equations are specified, Ihey can be combined into a malhematical model. The model can then be improved by comparison of Ihe simulalion 10 reality and correcting the model for a better fit. The Basic Equations for Ihe Social Model The parameters in all the equations except 5.4 and 5.6 were eSlimated by calibrating general equa- tions representing the decision processes for a flood control proposal. Differences are manifest in the val. ues of the paramelers which vary with the groups and stages for which the equations were calibrated. The dependent variable in the first basic equation is the evalualion of a flood control proposal by a particular agency or defmed population. General equation A general form of this equation is: Ygp = bO + biXIO + b2X20 + b3x30 + b4X40 + bSXSOXSl + b6X6Q'60 + b7X70z70 + baXSOz80 + b9XgOz90 + blOX lOOZlOO . (5.1) in which ygp predicted evaluation of a specific flood control proposal by a parti- cular group regression cons tan t b 10 = regression coefficien t factors from the population or agency factors from the flood control proposal = factor from other sources of in. fluence of llood control proposal evaluation bO bi. X z v The sub scripting procedure used in Equation 5.1 is designed to indicate the term number in which the variable appears and also its location wilhin the term. For example, in the term bSX50XSI' the sub- script 51 designates that the variable XSi appears in the sixth position on the right side of the equation (0 + 5) and thai it is Ihe record X-variable used in the term (0 + 1). The terms shown in Equation 5.1 arc simpli- fied. For instance, although only simple linear re- lations are expressed in the first three terms, data often contain complex nonlinear relationships which could be rellected in a computer model. Some of the necessary terms may be more complex than those shown in Eqnation 5.1. Oassificatian of terms in the general equations Four types of terms are included in Equation 5.1. Type I Terms are those in which only faclors describing the population or agency occur and no in- teraction occurs among them (such as bI XIO)' These terms represent factors in populations, agencies, or proposals which influence reaction to a flood control proposal but whose effect is independent of the mea- sured values for any other factors. This type of term would have the same influence on attitude toward a llood control proposal regardless of the proposal be- ing evaluated and may be considered as reflecting a tendency to accept or reject flood control proposals in general. Type II Terms are those in wh.ich more than one variable describing the population or agency occurs in the same term of the equation. Terms of this type happen when a variable has an effect on evaluation only when and to the extent that another variable is also present; in other words, a contingent relationship occurs such as if high educntion had an effect only if high income were 4.llS0 present. An example of this type is the term beginning with b4 in Equation 5.1. Variables from the agency or population which do not account for differences between groups in per- ceptions of particular flood control proposals can be combined in expression: bci = bO + b 1 XIO + b2X20 + b3X30 + b4X40X41 (5.2) It is apparent from Equation 5.2 thai the greater the contributions to the explanation of the dependent variable the less the value of the remaining bo term in Equation 5.1. Substituting 5.2 in 5.1 gives: y = be) + b4z40 + b6X60z60 + b7X70z70 + bSXgOzSO + b9X90z90 + blOXlOOvlOO . . (5.3) All terms on the right hand side of Equation 5.2 are constant for a particular proposal, but their values 60