I am having a hard time to understand an issue which has been pointed out to me, though with limited further explanation. I am testing an hypothesis about the pattern-dependence of a variable (e.g. something like conflict). That is, I believe that the degree of this specific type of conflict is likely to be high at time 1, but that if high at time 1 it will decrease at time two. I regressed the value at time 1 on the variable at time 2. I understand this is too simple, and I was told it is not correct to test this hypthesis. But not how to test it, unfortunately. Would be super grateful for any suggestion, possibly a simple way.

Thanks in advance