Bob Johnson

Michael doesn’t see the link between Summers’ resignation and the upcoming campaign, in part because Summer doesn’t play a campaign role, and because the public really doesn’t know who he is. The fact that Summers is policy wonk, however, is precisely my point – as a presidents turns from governing to campaigning, Summers expertise becomes less useful. And while it is true that not many people know who Summers is, the fact that Obama’s senior economic adviser is resigning certainly grabbed media attention. Moreover, it removes one potential point of controversy that will undoubtedly come up in November.

I don’t doubt, as Evan suggests, that Obama and his senior aides certainly calculated the political impact of Summers’ resignation, although I won’t necessarily go so far to believe he orchestrated this and previous resignations for political purposes heading into the midterm. Still, presidents have jettisoned key aides before who have become political liabilities (see Rumsfeld under Bush II.) I don’t think Summers was necessarily viewed as a liability but his resignation does provide Obama with an opening to signal his intent to change policy directions.

Finally, Bob raises a great question: when can we infer causation from correlation? The argument that Dunn Tenpas and I make is actually a bit more nuanced than the bare-bones presentation above, and relies on some qualitative evidence as well (we tell some stories about specific staff turnover during the Clinton presidency, for example). I don’t really do the argument full justice in this blog. But even so Bob raises a critical point. One response would be: can you suggest an alternative explanation for the correlation we present? Zach and Michael suggest a potential alternative: changing standards of media coverage that make serving in the White House less enjoyable. Dunn Tenpas and I link that changing media coverage to broader changes affecting how candidates run for office, so we don’t differ too much from their media-centered explanation. But let me turn it back to you Bob (and to other readers): what alternative explanation can you come up with?

In statistical analyses such as your second graph does the appearance of a relationship prove causality, or does there have to be a statistical “smoking gun”? Otherwise how do you rule out coincidence? It seems to me there could be other graphs substituting global warming or population growth for example that would yield the same correlation as the increase in delegates selected in primaries.

Or does it come down in the end to the judgment, or common sense, of the political scientist who is arguing that the relationship shown is indeed a causal one?

Bob Johnson