ESTIMATING STATUS AND TRENDS USING LTRMP SURVEY DATA

Brian R. Gray
U.S. Geological Survey, Upper Midwest Environmental Sciences Center, La Crosse, 
WI 54603

Reporting status and trends statistics is standard for most monitoring programs.  
Given a probabilistic design, these statistics are reasonably estimated using 
so-called ‘design-based’ and ‘model-assisted design-based’ estimators.  
Inferences using these methods, being based on the survey design rather than on 
subjective selection of a model, are arguably more standard and less vulnerable 
to challenge.  However, the use of design-based methods requires attention to 
monitoring design features.  For the LTRMP, these features include 
stratification (geomorphic and temporal), sampling weights (we don’t sample 
strata proportional to size), clustering (veg’s sites  and, for all components, 
years) and finite population correction factors (strata with small numbers of 
cells).  These methods have previously been used to generate status estimates 
but have not historically been used to estimate trends in LTRMP data.  
Estimation of status is described, and monotonic trend estimation for linear 
(water), dichotomous (relative frequency) and count data are illustrated using 
SAS® and SUDAAN® software.

Keywords: design-based estimates, LTRMP, status, survey regression, trends