We apply a dynamic spatial model to interest group ratings of the members o
f Congress over the period 1959-1981. Spatial distances between an interest
group and the members of Congress are assumed to be monotonic with the rat
ings. Our pooled cross-sectional time-series data set consists of 203,387 r
atings by 59 interest groups. We restrict the spatial coordinates of the in
terest groups and members of Congress to be polynomial functions of time. T
wo significant dimensions are recovered: the first dimension, which account
s for approximately 75% of the variance, represents liberal-conservative po
sitions on economic issues; the second dimension, which accounts for approx
imately an additional 5% of the variance, represents liberal-conservative p
ositions on social issues. Nearly all the interest groups and most members
of Congress are ideologically consistent. They are either liberal on both d
imensions or conservative on both.