Effects of Dialogue Meetings - Pre-analysis plan Phase 1

Sickness absence entails large individual and societal costs. Dialogue Meetings (DMs) where the absentee, the employer, and often the physician, discuss arrangements for full or partial work resumption have been in place in Norway since 2007. In collaboration with the Norwegian Labour and Welfare Administration (NAV) we are conducting a large scale randomized natural field experiment to investigate if and for whom DMs have effects. The experiment is still ongoing and is conducted in several phases to test for mechanisms iteratively. While the experiment is still ongoing, it is possible to analyze data on the individuals that have already been involved in the experiment so far (early sample). To do so we proceeded as follows: first, we randomly divided the early sample into an early training sample and an early test sample. NAV then merged the early training sample with outcome data and covariates and handed the data over to us. Based on the results in the early training sample, we here register some key decisions that will be followed in analyzing the rest of the data (the analysis sample, which comprises the late sample and the early test sample). The results from the early training sample are presented in a companion document and will be subject to a direct replication using the early test sample. Based on the findings in the early test sample we will decide whether and how to change the intervention. Such changes will be pre-registered at a later stage.