Rosenquist, Fowler and Christakis (2011) conducted a correlational study (they did not randomly allocate people to be friends). As we know, this means that cause-and-eect inferences cannot be made. However, their study was longitudinal and they looked at how the variables were changing dynamically. For example, they looked at how acquiring a new friend today correlates with developing depressive symptoms several years later. Does this change anything in terms of our ability to make cause-and-eect inferences?