Various approaches are currently used to interpret experimental data on fission. We critically examine a wide set of observables measured for fission $^{206,210}$Po nuclei at medium excitation energy, and illustrate the ambiguity in current analysis. Dynamical calculations based on the four-dimensional Langevin equation using a macroscopic potential energy landscape are performed, and found to consistently describe available measurements. This observation calls into question the robustness of recent analysis based on statistical-model calculations and concluding, on the contrary, to substantial shell effects at the fission saddle point $^{206-210}$Po. The inconsistency in interpretation reached by the two approaches shows that, depending on the system, the conclusion can be strongly model-dependent. Although this may not be surprising, it emphasizes the today still limited reliability of firmly extracting fundamental nuclear properties from customary analysis