Study on attitude towards regulated social activities have been carried out in many areas (such as tax and zakah payment). However, many of these studies applied a single score of attitude in their analyses. Such a procedure, to some researchers is considered less informative, especially in the study of a complex attitude which has several dimensions. Many researchers have suggested that attitude towards a complex object should be studied by decomposing the object or issue into smaller and less complex elements on the basis of component parts, specific functions, or particular contexts. Thus, this paper offers a comparative study of outcomes between attitude measured by a single summative score and attitude measured by multidimensional factor scores. The object of attitude in this paper is zakah on employment income by eligible Muslim. In the first approach, a total of 24 items of attitude were used to represent the single score of attitude. In the second approach, principal component analysis with varimax ratation was first applied to determine the underlying dimensions of attitude. Each dimension was then named and treated as anew variable, each measured by the factor scores. Both approach were applied separately to an analysis on compliance behavior of zakah on employment income. Results suggest that attitude measured by multidimensionality scores is more informative as compared to the single summative score. Futher, the use of multidimensional scores in multivariate logistic regression improved the goodness of fit of the model over that of the single score of attitude. Thus, this improvement affects the interpretation of the whole model with respect to the relationship between the independent variables and the dependent variable, which is zakah compliance.