The ICC is widely studied for discrete outcome variables. It is much less studied for so-called functional data, i.e., outcome measures that are functions, over time or some other domain. Reliability estimates for curve data have been reviewed and proposed Pini et al.(2022); Schelin et al. (2021), but there is a need for further studies.
We propose and evaluate methods to construct valid confidence bands for the ICC estimates based on simulated curves. These confidence bands offer a statistical tool to quantify the uncertainty associated with the estimated ICC values, consequently providing a range within which the true ICC is likely to fall.
Developing accurate and reliable confidence bands is crucial for researchers and practitioners in the field, as it facilitates decision-making based on the estimated reliability parameters.