Statistical models of high-latitude plasma convection have been used in a wide range of studies pertaining to the ionosphere and thermosphere, and they are beginning to be used in various space weather applications. However, the statistical convection models only provide average, not instantaneous, convection patterns, and it is not clear if they are real convection patterns or blurred images of convection. It is also unclear how reliable these convection models are for applications involving ionosphere-thermosphere specifications and forecasts. To address these issues, a quantitative analysis was conducted of the Weimer <2001> empirical convection model, which is the most comprehensive model of high-latitude convection that has been constructed to date. First, criteria were established to determine whether or not a modeled convection pattern was correct for a given set of geophysical conditions. The criteria adopted were reasonable but stringent. Then the cross-track ion drift velocities obtained from Weimer <2001> were compared with the corresponding velocities measured by the DMSP F13 satellite. The comparisons were done for nearly a year (1998) of satellite crossings of the northern polar region (4430 successive crossings). The results indicate that the Weimer <2001> model is able to produce the gross structure in the convection pattern, i.e., it is good in a statistical sense. However, it does not adequately capture the mesoscale spatial structure and convection magnitudes observed by the DMSP satellite. Typically, it was able to capture real (instantaneous) convection features in only about 6% of the satellite crossings. Although empirical convection patterns are invaluable for scientific applications, these results have important implications for using empirical convection models in space weather applications.