The sensitivity of derived estimates to the Measurement Quality Objectives for independent variables

The effect of varying the allowed measurement error for individual tree variables upon county estimates of gross cubic-foot volume was examined. Measurement Quality Objectives (MQOs) for three forest tree variables (biological identity, diameter, and height) used in individual tree gross cubic-foot volume equations were varied from the current USDA Forest Service Forest Inventory and Analysis specifications in a simulation under alternative error models. Assuming unbiased errors may lead to a different control strategy than assuming unbiased errors. Strengthening the MQO for diameter was shown to help reduce the overall variance of volume estimates if diameter errors are slightly biased. Height errors responded favorably to increased control under both the biased and unbiased models. County volume estimates are somewhat robust to the MQOs for biological identity. However, increased control of biological identity did play a more important role when the underlying distributions for diameter and height were assumed to be biased than when these errors were assumed to be unbiased.