The observed characteristics of radio auroral backscatter at large aspect angles have never been adequately explained by theories of plasma turbulence that otherwise seem to be satisfactory. Recently, the first nonlinear theory on the subject has appeared in the open literature. In this theory, under the action of a driving electric field, the maximum aspect angle depends on the combined contribution of frequency broadening due to mode coupling and anisotropic anomalous collisions caused by the interaction of electrons with the fluctuating electric fields in the waves. In this paper an attempt is made to test the theoretical predictions by comparing them to observations of mean frequency broadening during typical auroral electrojet conditions. The analysis was based on a large number of Doppler spectra compiled from several VHF radar backscatter experiments with magnetic aspect angles in the range from about 1 to 10 deg. The theory fails to explain echo occurrence at directions perpendicular to the E¿B flow, but otherwise it can account for most of the observations. The evidence shows that the effect on large aspect angle echoes of frequency broadening along is insufficient and that the key role in fitting the data is played by the parallel anomalous collision frequency &ngr;*e∥. The agreement between theory and experiment, however, is possible only for large enhancements of &ngr;*e∥ relative to the normal electron neutral collision frequency &ngr;e, with the increase ranging between 1 and 2 orders of magnitude. This imposes difficulties because the large &ngr;*e∥/&ngr;e ratios imply a pronounced ion acoustic plasma turbulence along B inside the electrojet layer. To our knowledge, there is no observational evidence to support this theoretical implication. ¿ American Geophysical Union 1994