Frequentist and Bayesian approaches for comparing interviewer variance components in two groups of survey interviewers