Skip to main content

Evaluating the Impact of Nonsampling Errors on Erosion Estimates for the Conservation Effects Assessment Project

Apr 12, 2013 - 12:00 PM
to , -

Evaluating the Impact of Nonsampling Errors on Erosion Estimates for the Conservation Effects Assessment Project

 

Date: Friday, April 12
Time: 12:00 pm -- 1:00 pm
Place: Snedecor 2113
Speaker: Andreea L. Eriulescu, Department of Statistics, ISU

Abstract:

The Conservation Effects Assessment Project (CEAP) is a series of surveys intended to evaluate environmental outcomes associated with conservation practices. Four sources of error in CEAP are nonresponse error, location error, frame problems, and processing error. Nonresponse error occurs due to refusals and can be evaluated using auxiliary data available for the full sample. Location error, resulting from imperfections in data collection protocols and GPS instruments, refers to differences between the sampled location and the location at which data are collected. Location error was judged to be problematic in early CEAP surveys, and the magnitude of this source of nonsampling error is expected to decrease as data collection technology improves. Constructing a frame that covers the whole population of interest and does not cover domains that are irrelevant for CEAP is difficult. Only farming operations with land in cropland or certain kinds of hayland or pasture are eligible for CEAP, but current information on landuse is not available at the sample design stage.  A fourth source of nonsampling error in CEAP arises because the original collected data (responses to survey questionnaires) are processed through a computer model called the APEX model. The APEX model calculates an erosion value and an erosion index for each respondent as functions of the survey responses and auxiliary information related to climate and soil characteristics. Imperfections in the APEX model may lead error in CEAP estimates.  Possible ways to evaluate effects of nonresponse error and error due to frame inefficiency on small domain estimates based on the CEAP survey data will be discussed. Limitations in our abiblity to evaluate location error and error in the APEX model will also be considered.