Data and scripts from: Design and use of monitoring networks: Few-large versus many-small (FLvMS) and multi-scale analysis

In order to learn about broad scale ecological patterns, data from large-scale surveys must allow us to either estimate the correlations between the environment and an outcome of interest and/or accurately predict ecological patterns. An important part of data collection is the sampling effort used to collect observations, which we decompose into two quantities: the number of observations or plots ($n$) and the per-observation/plot effort ($E$) (e.g. area per plot). If we want to understand the relationships between predictors and a response variable, then lower model parameter uncertainty is desirable. If the goal is higher predictive performance, then lower prediction error is preferable. We aim to learn if and when aggregating data can help attain these goals. We examine the impacts of aggregating observations for count and continuous data. Through simulations, we generate data and fit models at different degrees (e.g. groups of 10, 60) and types of aggregation, and examine parameter uncertainty as well as prediction error. We compare the findings from simulated data to real data in an application to tree density of selected species from Forest Inventory and Analysis (FIA) data. In particular, we fit models to FIA data that have been aggregated via distance and covariate similarity or US EPA ecoregions.