State dependence in immunization and the role of discouragement

We investigate whether having a child immunized at a prior schedule genuinely increases the likelihood of vaccinating the child at the subsequent schedule. We use longitudinal data from the Growing Up in New Zealand study and apply a dynamic random-effects model that also controls for the initial immunization status. Prior to any covariate-adjusted estimations, our data shows that almost 96% of the children immunized at the previous schedule are also immunized at the subsequent schedule. In comparison, only 29% of children who were not immunized at the prior schedule receive immunization at the next milestone, thereby indicating an unadjusted state dependence in immunization of 67 percentage points (p.p.). Upon controlling for relevant covariates and unobserved heterogeneities, the genuine state dependence in immunization is, on average, estimated to be 20 p.p. Importantly, the magnitude of the state dependence is greater for Māori (by 5 p.p.) and also greater for mothers that report being discouraged from having their child immunized during the antenatal period (by 10 p.p.).