We compare two methods for estimation of path-component spreads in bi-azimuth and delay, namely the clustering approach and the density approach, in indoor environments. Monte Carlo simulations reveal possible shortcomings of the clustering approach. Published estimates of component spreads in delay, azimuth of departure and azimuth of arrival obtained with the clustering approach are reviewed and contrasted with estimates gathered using the density approach. A detailed study of these experimental data, aided with the insight gained from the simulation results, leads to the conjecture that in some cases the estimated spreads computed with the clustering approach are too large. The settings of the path component spreads of two widely used models including bi-azimuth delay dispersion, namely the Winner II Model and the 802.11 Tgn Model are revisited based on these findings. The investigations stress the obvious, but apparently sometimes forgotten, importance of validating the behavior and performance of channel parameter estimation techniques before using these tools to extract empirical information from measurement data.