Dynamic Time-to-Event Models for Future Call Attempts Required Until Interview or Refusal

The rising cost of survey data collection is an ongoing concern. Cost predictions can be used to make more informed decisions about allocating resources efficiently during data collection. However, telephone surveys typically do not provide a direct measure of case-level costs. As an alternative, I propose using the number of call attempts as a proxy cost indicator. Most previous studies have focused on logistic regression and linear regression for cost estimation. This study advocates time-to-event models for predicting survey costs. To improve cost predictions, I dynamically adjust predictive models for future call attempts required until an interview or refusal during the nonresponse follow up. This update is achieved by fitting models on the training set for cases that are still unresolved as of a particular number of call attempts. This approach accommodates additional paradata collected on each case up to that point. I use data from the Health and Retirement Study to evaluate the ability of alternative models to predict the number of future call attempts required until interview or refusal. These models include a baseline model with only time-invariant covariates (discrete time hazard regression), accelerated failure time regression, survival trees, and Bayesian additive regression trees within the framework of accelerated failure time models.