I don't think so. My colleague may.
Again, this is a cross-sectional survey. Typically, cross-sectional surveys are good for association. We know that the people who answered their employment status as such-and-such and their answers to disorders are such-and-such, so we can do the graph that you have in front of you. Causality really requires, as I said, going back to people so that we know what comes first and then we know the transition. If someone was in a given state of employment and—