State dependence in immunization and the role of discouraging information

Does having a child immunised at the prior schedule genuinely impact the likelihood of vaccinating the child at the following schedule? Using longitudinal data from Growing Up in New Zealand study, we apply a random-effects probit model that also controls for the initial immunisation status. We detect sizeable state dependence in immunisation, indicating that the likelihood of a child increases, on average, by 21 percentage points if the child was immunised at the previous schedule compared to if not. This effect is further exacerbated if the mother received antenatal discouraging information on immunisation.