Abstract
In natural resource management, high-quality monitoring data is necessary to make informed management decisions. Multiple observers collecting data for a monitoring program can potentially introduce a large amount of variance decreasing data quality. In this thesis, an application of mixed-effects models with relaxed homogeneous variance assumptions (called heterogeneous-variance mixed-effects models) was introduced for quantifying observer variance in natural resource monitoring programs with large spatial and temporal extents. A national-level land-health monitoring program was used to describe and demonstrate the method. Two example analyses identified differences in observer variance across regions and years. These examples illustrated several potential uses of heterogeneous-variance mixed-effects models for evaluating influences on monitoring data quality, determining appropriate use of monitoring data, and suggesting potential strategies to reduce observer variance in future monitoring data collection efforts. Data requirements, extensions, and other applications for heterogeneous-variance mixed-effects models for evaluating observer variance in natural resource monitoring programs were also discussed.