Any author would welcome such an informed and constructive comment as Burt makes on my ‘New Spatial Theory of Party Competition’. I have just one minor qualification, and a suggested extension to his remarks.

The qualification regards his uncharacteristically reproachful remark about the absence of ‘true success rates’ (p. 652) – the fact that in the original article I did not test each model over all the data. This criticism really derives not from the inappropriateness of my statistics but from the fact that my purposes were not his purposes. I was not concerned to see how well one model would perform over all the data but rather to sort out which model would apply best to which set of parties. As my starting point was the idea that different parties would adopt different decision rules under uncertainty, this procedure was one which suited my purposes, though not his.

My suggested extension concerns the demonstration that parties actually alternate less than might be expected by chance, ‘suggesting that there may be a slight tendency for continuation of policy movements’ (p. 657). Clearly this casts doubt on the validity of the original decision rule, which postulated that (under uncertainty) parties reversed previous policy at each election.