We present a systematic analysis of point-contact Andreev reflection (PCAR) spectra for ferromagnetic materials, using both modeling and experimental data. We emphasize the importance of consistent data analysis to avoid possible misinterpretation of the data. We consider the relationship between ballistic and diffusive transport, the effect of different transport regimes on spin polarization measurements, and the importance of unambiguous identification of the type of transport regime. We find that in a realistic parameter range, the analysis of PCAR spectra of purely diffusive character by a ballistic model yield approximately the same (within similar to3%) values of the spin polarization and the barrier strength Z larger by similar to0.5-0.6. We also consider the dependence of polarization values on Z, and have shown by simple modeling that letting the superconducting gap vary as an adjustable parameter can result in a spurious dependence of the spin-polarization P-c on Z. At the same time we analyzed the effects of finite Z on the apparent value of P-c measured by the PCAR technique, using a large number of examples from both our own measurements and from the literature. We conclude that there is a system-dependent variation in P-c (Z), presumably due to spin-flip scattering at the interface. However, the exact type of this dependence is hard to determine with any statistical certainty.