This note challenges the use of census data for evaluating changes in
commuting speeds, partly because these data require indirect distance
measures, but primarily because they do not differentiate between non-
stop trips and trip chains. The proportion of trip chains to total tri
ps increased between 1983 and 1990 (Nationwide Personal Transportation
Survey). Hence, results of declining speeds based on census data may
be spurious. When trips are broken down by type using Nationwide Perso
nal Transportation Survey data, speeds increased between 1983 and 1990
for all categories, in all but one of eight cases studied, by signifi
cant amounts. (C) 1997 Elsevier Science Ltd.