Jitter has long been used to describe period instability in voiced speech signals. In spite of this long history of measuring and modeling jitter, the ways the different glottal pulse phases are affected by jitter differ across studies. The models have quite dissimilar implications, and their selection has been rather arbitrary in the literature. This paper describes different choices for modeling jitter within the glottal pulse, and their implications in several speech processing tasks. Based on the discussion presented, any model selection departs from some accredited standpoint in voiced speech processing. Experiments to evaluate the effect of different models on the performance of periodicity perturbation measures are carried out on synthetic signals. The results obtained show large differences across models, demonstrating the need to evaluate each model's fit to actual data.