Dear Mikidadi.

You ask the following: Which is the best way to analyze causality in a behavioral  study? 

F. B. Skinner, the most famous of all behaviorists, once said that the variables of which human behavior is a function lie in the physical and social environment. So, to analyze causality in a behavioral should we could, for example, manipulate a certain feature of one's physical (e.g., the color of classrooms in an elementary school) and/or social environment (e.g., teaching styles) and see whether this or theses change(s) impact on pupils' learning, such as assessed, for instance, by pupils' grades or marks in a test situation.  

Let me add that the procedure or way to analyze causality in a behavioral study is not much different from analyzing causality in a cognitive study. In this case, we could, for example, introduce some changes in pupils' cognitive style (e.g., field dependency/independence) and see whether this chance affects pupils' performance, for instance, in a Piagetian liquids conservation task.  

This means that in both cases we manipulate the independent variable and looks at its effects on the dependent variable. Of course, in both cases, it would be necessary to have an experimental group and a control group.

I hole I has not misinterpreted your question and that this is of some help for you.

Best regards.

For behavior in the economy there is a classic "Causality in Economics" by John R. Hicks, England