Using fixed effects panel analysis to investigate correlations between civic membership and political engagement: causation or self-selection?