Theory-based modeling is a way of modeling in which qualitative data, namely linguistic expressions, are participating for making a model representing a real system. Modeling in Social Sciences usually uses quantitative data, out of the questionnaires, in addition to the qualitative data. Structural Equation Modeling (SEM) is one of the new, complicated and widely used methods toward making models for different social aspects of research, in which the desirable model is the one which satisfies both statistical and theoretical indices. SEM has an algorithm that includes five recursive steps in which the quantitative and qualitative aspects are being checked one after another regarding their corresponding conditions including statistical indices and theoretical basis respectively. The recursive process of finding a desirable model is only terminated when the both mentioned conditions are satisfied, although it is possible not to find a good model that complies our demands, in such cases more theoretical researches are necessary towards either adding or removing other parameters into our default model. In addition to the SEM, the familiar Fuzzy ways of modeling, Set Analysis (SA) and Cognitive Maps (CM) are considered as some effective qualitative ways of modeling in Social Sciences. In SEM and CM the inner causal relationships among the variables, inside our model, are analyzed therefore the inner structure of a system is presentable for a researcher to find a better insight from the inside of the model. The usual Fuzzy Models like Fuzzy Inference Systems (FIS) and Anfis have input-output point of view, so they are mainly used for prediction or simulation applications. SA is confronting two different variables in its crosstabs in which corresponding levels of both variables are met and out of a process corresponding linguistic expressions are made. If Fuzzy Logic were used, that would be possible to make a set of relatively true linguistic expressions too. Generally all different ways of modeling that use qualitative data, alone or in addition to the quantitative ones, are useful in applications in which quantitative data are not reliable enough to make a certain decision. Socialists in their social researches do not usually rely on quantitative results because these digits are seemed vague from the roots of finding in their points of view, but the use of qualitative way of modeling in industrial applications is not ignorable either. “Desirable regulation of the output in a semi-automation process on the basis of its user’s demand” would be a tough selection, if we considered the relative meaning of desirability and different users’ demands. The semi-automation is necessary if the user’s demand is supposed to be a part of the close-loop control structure, achieving desirabl Keywords : structural equation modeling, set analysis, pure fuzzy inference system, cognitive map.