The first four chapters in part III focus on hunting causes in economics; the last on using them. None, I am afraid, even starts on the job I urge to be so important: showing how our methods for hunting causes can combine with other kinds of knowledge to warrant the uses to which we want to put our causal claims.

I look at two different methods employed in economics for hunting causes. The first uses econometric techniques, the second, theoretical models. The econometric techniques discussed are a caricature of what happens in economics. I look at only one author who claims to get causes from probabilities – Herbert Simon – and then only at the simplest imaginable cases. That is because I think Simon has a clear idea of the difference between selecting a model that is accurate about the underlying probability from which the data are drawn versus one that is accurate in describing causal relations. I want to focus on this difference without intertwining it with any of the pressing problems that must be treated in trying to ensure correct estimation of probabilities.

The two chapters on Simon here might be seen as part of a trilogy: Simon as interpreted by me, where I suppose that the trick in getting causes from probabilities depends on the characteristics of the ‘exogenous’ variables; Simon as interpreted by Kevin Hoover, who supposes that what matters are the parameters that describe the quantities we can directly control; and Simon as interpreted by Damien Fennell, who takes the idea of separate mechanisms to be central.