B. Survey Design and Sampling
1. Survey Design
For this Export Cargo Survey, the target population was commodity sorghum from the nine key U.S. sorghumproducing states representing more than 98% of U.S. sorghum exports. A proportionate stratified sampling technique was used to ensure a sound statistical sampling of U.S. sorghum exports. Two key characteristics define the sampling technique for this report: the stratifcation of the population to be sampled and the sampling proportion per subpopulation or stratum.
Stratifcation involves dividing the survey population of interest into subpopulations called strata. For the Export Cargo Survey, the key sorghum-exporting areas in the United States are divided into two geographical groupings, which we refer to as EOs. These EOs are identified by the two major pathways to export markets:
- The Texas EO includes export terminals along the Texas Gulf Coast, primarily League City (Houston Area) and Corpus Christi; and
- The NOLA EO comprises the export terminals near the Mississippi River Delta.
To determine the sampling proportion of each EO, the Council used historical and projected data from USDA’s National Agricultural Statistics Service (NASS), World Agricultural Supply and Demand Estimates (WASDE), and Export Grain Information Service (EGIS), along with private sources, to estimate the proportion of 2015/2016 sorghum exports from each EO. The sampling proportion (each EO’s proportionate share of the total estimated foreign exports) ultimately determined the number of sorghum samples to be collected from each EO. The specified sampling proportion for each EO are as follows: NOLA EO – 21%; and Texas EO – 79%.
The number of samples collected from each EO was established in order for the Council to estimate the true averages of the various quality factors with a specifc level of precision. The level of precision chosen for the Export Cargo Survey was a Relative ME of no greater than ± 10%, which is a reasonable target for biological data such as these sorghum quality factors.
To determine the number of samples for the targeted Relative ME, ideally the population variance (i.e., variability of the quality factor in the sorghum exports) for each of the quality factors should be used. The more variation among the levels or values of a quality factor, the more samples required to estimate the true mean within a given confidence level. In addition, the variances of the quality factors typically differ from one another. As a result, different sample sizes for each of the quality factors would be needed for the same level of precision.
When population variances are not known, variance estimates from similar data sets are used. Although a reliable source of chemical composition and physical factor data was not available, variances and Relative MEs for the grade factors were calculated using the EGIS sorghum export data, and were used as proxies. Based on these data, a total sample size of 167 would allow the Council to estimate the true averages of the grade factor characteristics with the desired level of precision for the U.S. Export Aggregate. Applying the sampling proportions previously defined to the total of 167 samples resulted in the following number of targeted samples from each EO: NOLA EO – 35 samples; and Texas EO – 132 samples.
2. Sampling
The sampling was administered by FGIS as part of their inspection services. At the time of this survey’s approval in September 2015, new crop sorghum was already being loaded at export points. Therefore, it was decided to start the sampling period as soon as possible. FGIS sent instruction letters to the Texas and NOLA field offces, and the sampling period began the first week of September. The FGIS field offce in League City, Texas was responsible for overseeing sample collection in the Texas EO, and the FGIS field offce in New Orleans, Louisiana was responsible for the oversight of sample collection in the NOLA EO.
Representative sublot samples from the ports in Texas and NOLA were collected as ships were loaded. Samples for grading are obtained by a diverter sampling device approved by FGIS. The diverter sampler “cuts” (or diverts) a representative portion at periodic intervals from a moving stream of sorghum. A cut occurs every few seconds, or about every 500 bushels (about 12.7 metric tons), as the grain is being assembled for export. The frequency is regulated by an electric timer controlled by offcial inspection personnel, who regularly ensure that the mechanical sampler is functioning properly.
While the sampling process is continuous throughout loading, a shipment or “lot” of sorghum is divided into “sublots” for the purpose of determining uniformity of quality. Sublot size is based on the hourly loading rate of the elevator and the capacity of the vessel being loaded. Sublot sizes range from 35,000 to 75,000 bushels. All sublot samples are inspected to ensure the entire shipment is uniform in quality.
The sampling frequency for each EO was identical: sublots with identifcation numbers ending in 0, 3, 5, 7, and 9 from each lot were sampled. Since quantitative aflatoxin testing is not required for exported sorghum shipments, the survey’s sampling protocol did not require a sublot to have aflatoxin testing conducted in order to be sampled.
For each sample, the FGIS field staff collected a minimum of 2500 grams. The samples were congregated at the field offces and mailed to the Texas A&M Cereal Quality Laboratory (CQL). Refer to the “Testing Analysis Methods” section for the description of the testing methods employed in the study.
The sampling period ended when the targeted number of samples per EO was reached, which occurred on November 6, 2015 for the Texas EO and on September 18, 2015 for the NOLA EO.