•In this model, the subject effect was considered a random effect, and time was a covariate. As both the protocol effect and the protocol x time interaction were not significant, the best model for the neck was simply an overall intercept ( - 8.11) and slope (5.77) for all protocols combined.