I'm in the midst of this kind of study and would like to use Structural Equation Modeling to assess the complex relationships among many variables related to Bystander Intervention.
I think it should be assessed in accordance with your hypothesis. Both exogenous and endogenous variables to estimate causalities you want to reveal, maybe??
I have been personally in a bystander situation and in retrospect realized that my ascribing to social norms of wanting to maintain a harmonous environment among acquaintances and my uncertainty that what I thought was going on was actually going on kept me from speaking out in the group when both offender and possible future victim were there.
to do this type of study well if there are no previous studies that can help identify variables might be to find individuals who participate in high risk situations and interview a few (qualitative) to help identify variables. Or perhaps a survey with open ended questions on Amazon Mechanical Turk https://www.mturk.com/mturk/welcome