I am flattered that a political scientist of William Keech's caliber uses words such as ingenious and outstanding to describe my work. I am also grateful that he took the time and was given the opportunity by the Review to show the implications of my theory of political support for a stylized case where unmeasured sources of support for British governments among the occupational classes are nil (aqj = 0), the only performance variable valued by voters is the unemployment rate, which has a reasonable contemporaneous impact of –0.02, and where the decay rate parameter g for past performance outcomes is set equal to the sensible value of 0.8 (implying a backward-looking discount rate of 0.25 per quarter).

A check of selected entries in Keech's Table 1 indicates they are correct, and I have no important disagreements with his interpretations of their behavioral implications. One caveat, however, about Keech's statement that my model “presumes that the performance of the old government affects the support of the new one with exactly the same slope (bn = b0 in equation A), but with opposite sign”: the absolute value of the partial derivative of Y* with respect to Zt–k in my model is b gk, which of course depends on lag k. b(bn, b0) is a scale parameter; the slope, as defined in the usual way, is given by the (time varying) derivative above.