|
Abstract:
|
Many factors influence the choice of an estimator for a given population parameter. One critical factor in that choice is the kind of survey data available. Farm operations often contract for specialized services they choose not to perform themselves. These specialized services are often negotiated on a per acre basis. The population parameter of interest for such services is the mean rate, defined as the ratio of total expenses for the service to the associated total acres. Estimating these mean rates is the goal of the Custom Rates Survey. The conventional estimation approach would be to estimate the total expenses and the total associated acres, and then to use a ratio estimate for the mean rate. However, asking for a farm operation's expenses for custom rates and the associated acres results in high survey non-response rates; hence, the Custom Rates Survey only asks for the corresponding rate. The conventional Horvitz-Thompson estimator based on the reported rates is a biased estimator for the mean rate as, defined above. An alternative approach, which may lead to an unbiased estimate, is to use a weighted mean of the reported rates.
|
