FLOOD09697 - Laserfiche WebLink

Home Browse Search

The "Proportion of Single Unit Structures in the City Block," may be important because of social interaction effects. An individual is influenced by the opinions of his associates and neighbors, especially those whom he feels share his situation. Therefore, a single residence occupant in an area of apartment type multiple dwellings24 can be hypothesized to be more likeiy to be positive on a flood control issue than single residence inhabitanls in a neighborhood dominated by singie residents. The first term related positively to the depend- ent variables is Term 2 (variable X2), "Proximity of Flood Experiences to Present Residence." If a person has actually experienced flooding and particularly in the place in which he lives, he is expected to be more likely to favor a flood control proposal. Variable X4, rrWillingness to Follow Government Agencies,11 would also be expected to be positively related to support for flood conlrol since most flood control proposals (all those considered in calibration of the model) are government recommendations. Term 5 or variable XS' "Knowledge of Local Flood Control Proposals," turned out to be positively related to the evalualion of flood control proposals for the cases tested but may not be generalizable to other areas or circumstances.25 The positive relation- ship in Equation 5.24 may be because the strong ef- fect of negalive factors on the evalualion of a proposal is largely accounled for by the acceptance functions. A positive relalionship would be reduced to the degree that the extra knowledge centers on negative factors. The sign and value of the coefficienl of Xs needs fur- ther testing by calibrating the general equation for each population to which the model is applied.26 The population variables may be much more im- portant in predicting differences in the reactions of dif- ferent populations to the same proposal than in the prc- sent application of predicling differences within the same population to different proposals. This is be- cause a variable must fluctuate to be studied. Conse- quently, data from a homogenous population may cause population variables which are important deter. minanls to be underrated. The matter can only be re- solved by applying the general equation to substan- tively different populations. 24Condominiums may be an exception because they are owned by the occupant. 25There IS about one chance in fourteen that the posi- tive relationship is spurious for the sampled population. This is the weakest significance level of terms in the equation. 26It may be impossible to predict the sign and magni- tude of Xs from the social and hydrologic conditions oran area. This interesting and worthwhile task, however, would be a major undertaking. Interaction terms The interaction terms in the Equation 5.24 are the heart of this equation and lhis model. They link the separate equations together and also account for most of the variance in the evaluation equations. In fact, the general equation could be calibrated using the interaction terms alone, and most of the variance accounted for by Equation 5.24 would be included. An indication of the approximate relative im- portance of the terms of the equation can be seen from the sizes of the coefficients in standardized form (see Chapter II). Allhough the coefficients are large for the interaclion terms, the r2 without the single vari- able terms would almost certainly be larger than that estimated from the ratio of the sum of the coefficients of the standardized interaction lerms to that of the sum of the standardized single variable terms. This is because of variance which would be explained by other variables if a term were removed. The measure- ment error in some cases is greater for variables in the interaction terms; and this would results in underesti- mation of the real effect of terms of which those vari- ables are part. In all the interaction terms the scale score was reduced by the number of items in the scale (Appen- dix C) in order to make the minimum Score equal zero. There is an implied assumption in lhis proce- dure that all attitudes are unidirectional in effect. Set- ting the zero poinl at one end of the scale means that there can be no opposites In sign. This means that the directional effect of an interaction with another vari- able depends entirely on the magnitude of the other variable. Unisigned variables are consistent with the idea of importance functions for which function scale scores were used in the interaction terms in the mathematical model. Importance normally is thought of as more or less, not negative or positive. In most cases, as applied in this model, this is reasonable.27 It may not be tenable for the Type IV term for the effecl of a revised decision agency evaluation on the public's attitude. The quantity three was subtracted from the per- ceived judgment score used in each of the Type III terms, called acceptance functions, so that "neutral" received a value of zero. This was done because of the substantive meaning of these interaction terms. The effect of a positive attitude about recreation, for example, would be the reverse if the proposal would lessen the quality or quantity2 8 of recreation than if 27It is assumed that almost no one will be against eco- logy, aesthetics, recreation, or effectivenss of a flood control proposal. 28Quality and quantity were not distinguished in ask- ing the perception of the effect of a proposal. This may be a distinction for some variables such as recreation. 75