How accurate are delegate reports that easily lend themselves to an exaggeration of the delegates’ roles? The best candidate for a reliability check is a straightforward behavioral question on constituent contacting, which we asked of congress delegates and a probability sample of the ordinary mass public in a subsample of voting districts in twenty-three surveyed townships. We asked township congress delegates the first of the four questions below and villagers the three that follow:

To construct a constituent contacting measure from delegate responses, I transformed response categories into values with 0 and 30 as extreme values and midpoints for the middle four categories. I then computed yearly averages, taking into account the lapse of time between the beginning of the delegate's term and the survey. On average, in the twenty-six surveyed townships, township delegates with no concurrent congress seat report 9.8 such contacts per year (or 13.6 per thousand constituents). In the twenty-three townships in which we also conducted surveys of villagers, the average is 9.6 contacts per year.

Three percent of villagers surveyed report contacting an official. Of these, 72 percent report contacting a township congress delegate, and 24 percent report contacting a county congress delegate. Of the 63 percent who recall when the contact was made, 85 percent report it was the year of the survey or the year or two before it.