Variable Weights

Information on Variable Weights

Not all variables are given the same importance in the filter, which is reflected in weights given to variables and to the four categories into which they fall – environmental, economic, social and cultural. The four categories are assigned weights that sum to 1, as are the variables in each category. Each variable, quantitative as well as qualitative, is converted into a value on a scale from 1-5 and then multiplied by the relevant weights. The result for each road is a composite “score” between 1-5, with higher numbers indicating great risk of environment, economic, social and cultural failure.

The Roads Filter weights can be changed by any user. In the original version posted by CSF, weights were mainly based on the following four criteria:

i. Certainty: When the data varies less over time, a higher weight is assigned to the variable. A geographical variable, such as topography, for example, will remain constant, while a social one, such as local objection to the project, is dynamic by nature.

ii. Type of data considered in scoring the variables: The Roads Filter’s economic and environmental variables are based on quantitative information, while its social variables depend on qualitative, sometimes subjective information. The cultural variables were based on quantitative information, but, because of incomplete information for the two variables analyzed in this group, a lower weight has been assigned.

iii. Importance of the variable as a determinant of the efficiency of the project for society: This score was determined based on a review of literature regarding the social evaluation of roads: (Reid, 2009; Fleck, 2009; Killeen, 2007; Dourojeanni, 2006; Fearnside, 2005; World Bank, 1998; among others).

iv. The number of variables considered in each category: The economic category considers five variables, the environmental category considers seven, while the social and cultural categories consider three and two variables, respectively.

Return to the Roads Filter