V. SURVEY AND STATISTICAL ANALYSIS METHODS
A. Overview
The key points for the survey design and sampling and statistical analysis for this 2015/2016 Export Cargo Report are as follows:
- Following the process developed for the previous Export Cargo Reports, samples were stratified according to Export Catchment Areas (ECAs) – the Gulf, Pacific Northwest and Southern Rail.
- To achieve no more than a ± 10% relative margin of error (Relative ME) for the U.S. Aggregate level and to ensure proportional sampling from each ECA, the targeted number of total samples was 430 samples, to be collected from the ECAs as follows: 272 from the Gulf, 92 from the Pacific Northwest, and 66 from the Southern Rail.
- Southern Rail ECA samples were provided by any of several official agencies designated by USDA’s Federal Grain Inspection Service (FGIS) that inspect and grade rail shipments of corn destined for export to Mexico. Gulf and Pacific Northwest samples were collected by FGIS field offices at ports in the respective ECAs.
- Export inspections of shipments from the Southern Rail ECA (interior) did not generate the targeted number of samples for this report within the sampling timeframe required to publish the report in a timely manner. As a result, 44 samples were collected for the Southern Rail ECA. Nonetheless, the U.S. Aggregate averages for the quality factors were weighted according to the targeted proportion by ECA.
- To evaluate the statistical validity of the number of samples surveyed, the Relative ME was calculated for each of the quality attributes at the U.S. Aggregate and the three ECA levels. The Relative MEs for the quality factor results were less than ± 10% except for three attributes from the Pacific Northwest and Southern Rail ECAs – total damage, stress cracks, and stress crack index.