“Attribute acceptance sampling.” Attribute acceptance sampling is typically used for evaluating a contractor’s internal controls. This includes the evaluation of policies, procedures, and practices to determine the adequacy of internal controls for detecting and preventing operational deficiencies. Since perfection is seldom expected, there is some level of non-compliance that can be tolerated. Attribute acceptance sampling is designed to discern whether non-compliance is within tolerable limits. Attribute acceptance sampling is not designed to identify questionable costs or the reasonableness of prices included in a contractor’s catalog. For purposes of DoD EMALL contracting personnel will use variable sampling as described below to support price reasonableness determinations.
“Catalog price.” A catalog price means a price included in a catalog, price list, schedule, or other form that is regularly maintained by the manufacturer or vendor, is either published or otherwise available for inspection by customers, and states prices at which sales are currently, or were last, made to a significant numbers of buyers constituting the general public (FAR 2.101).
“Commercial catalog.” A commercial catalog is a catalog of items meeting the FAR 2.101 definition of commercial item.
“Confidence level.” Confidence level is the assurance (or probability) that the amount being estimated by the sample will fall within a specified range (or interval) determined from sample results. A confidence interval is commonly (but not always) defined as the point estimate plus or minus the precision amount. For example, a 95 percent confidence level indicates that with repeated sampling under the same sampling plan, 95 times out of 100 the actual universe is expected to be within the interval computed from the sample results. In practical terms, this means that any single sample has a 95 percent chance of producing an interval that includes the actual universe amount. For a given sample size, the more confident an evaluator wants to be that the confidence interval contains the true amount, the wider the interval must be.
"Cost or pricing data" also encompasses decrement factor information.
"Decrement factor information" is the historical data necessary to determine the average difference between vendors' and subcontractors' proposed prices and the actual prices negotiated by the contractor with a specific supplier, all suppliers, or suppliers for a specific contract, commodity, or commodity group.
“Dollar unit sampling (DUS).” Dollar unit sampling is a substitute for stratification by dollar amount. In general, the two approaches are roughly similar in what they can accomplish. DUS does have an advantage in dealing with selected items that prove to be clusters of smaller physical units. Dollar unit sampling’s selection probability proportional to size (PPS) feature concentrates the sampling evaluation towards larger items much the same as stratification does for physical unit sampling. Collectively, the dollars making up an item give that item a chance of selection proportionate to its size in the universe. Dollar interval selection is used to select DUS samples. Dollar unit sampling implies that “dollar units” or “dollar hits” as opposed to physical units are being sampled. Physical units (e.g., invoices or price lists) are the sampling unit, with the sample items being identified by the dollar hits. In order to evaluate a dollar hit the item (e.g., the cost of the physical unit) containing the dollar hit must be analyzed.
Exclusive Dealers: When Original Equipment Manufacturers (OEMs) use exclusive distributors/dealers to sell their products, the Government usually must buy these products directly from the exclusive distributor/dealer. In these situations, the exclusive dealers are functioning as prime contractors, and the OEMs as subcontractors.
“EZ Quant.” EZ-Quant is statistical software developed and used by the Defense Contract Audit Agency (DCAA) and other Government agencies in evaluation of large universes of data such as a contractor bill of material. DLA contracting personnel will use EZ-Quant in performance of statistical sampling for DoD EMALL catalogs because of its wide acceptance within both the government and contractor community.
“Inferential or inductive statistics.” Inferential or inductive statistics are methods of using sample data taken from a statistical population to make actual decisions, predictions, and generalizations related to an area of interest. For example, in contract pricing we use stratified sampling of a proposed bill of materials to infer the degree it is overpriced or under-priced.
“Precision.” The term “precision” pertains to the amount or degree of probable error associated with an estimate (or the extent to which the sample findings may differ from the actual universe values or conditions). It measures the accuracy of a point estimate by showing, for a specified confidence level, how much the point estimate may vary from the true universe amount. In sampling for variables, precision can be expressed as either: (1) an interval about the point estimate obtained from the sample; or (2) a maximum or upper limit such as less than $25 or less than 5 percent error. In most cases the primary consideration influencing the evaluator’s selection of a desired level of precision will be the potential effect of the error on Government contract costs.
“Price Reasonableness Codes” (PRCs) are two digit codes comprised of a “Reviewer” code to identify the functional specialist(s) performing/participating in the price review; followed by a “Type Analysis” code to distinguish the nature of the price or cost analysis performed in support of the contracting officer’s price reasonableness determination (see PGI 15.406-3(a)(11)).
“Desired precision amount.” Desired precision amount is the amount of sampling error, stated as a dollar or percentage amount that is considered acceptable by the evaluator.
“Population.” A population is the set of all possible observations of a phenomenon with which we are concerned. For example, all the part numbers in an offeror’s commercial catalog would constitute a population. A numerical characteristic of a population is called a parameter.
“Random selection.” Random selection is a key principle in statistical sampling. To select randomly is to eliminate personal bias or subjective considerations (which cannot be expressed numerically) from the choice of a sample. Random sampling is a selection process in which each item in a stratum has a known probability (chance) of being selected.
“Sample.” A sample is a subset of the population of interest that is selected in order to make some inference about the whole population. For example, part numbers randomly selected from an offeror’s commercial catalog would constitute a sample. A numerical characteristic of a sample is called a statistic.
“Sampling frame.” A sampling frame is the physical (or electronic) representation of the sampling units from which the sample is actually selected. In sampling for variables, examples of sampling frames include a computer listing of a consolidated bill of materials, or a file of vouchers. For these sampling frames, possible sampling units are a part number, or physical voucher respectively.
“Significance level.” The significance level is equal to 1.00 minus the confidence level. For example, if the confidence level is 95 percent, the significance level is 5 percent. The significance level is then the area outside the interval which is likely to contain the population mean. Setting the significance level depends on the amount of risk you are willing to accept that the confidence interval does not include the true population mean. As the amount of risk that you are willing to accept decreases, the confidence interval will increase. In other words, to be surer that the true population mean is included in the interval, you must widen the interval.
“Statistical reliability.” Statistical reliability of sample findings is measured by two interrelated parameters, precision and confidence level. The evaluator must establish desired values of these parameters for either approach (physical unit or dollar unit) for variable sampling.
“Statistics.” Statistics is a science which involves collecting, summarizing, analyzing, and interpreting data in order to facilitate the decision making process. Statistical sampling is the preferred method of evaluation of less than 100 percent review of universe data. Statistical sampling is preferred over non-statistical (judgmental) sampling because of its advantages, which include objectivity, overall defensibility, and measurability of risk of substantial (or material) sampling error.
“Stratification.” Stratification is the partitioning of the evaluation universe into smaller groups according to a scheme that suits evaluation purposes. The evaluation universe consists of all of the transactions or other basic items within the scope of the evaluation. Stratification does not change the evaluation universe. Stratification is primarily used in variable sampling, and is rarely used in attribute sampling. The usual purpose of stratification is to enhance sampling precision, and thereby decrease the amount of evaluator time required to obtain adequate support for the evaluator’s conclusions.
“Universe.” A “universe” is a group of items or transactions from which information is desired. For purposes of PGI 15.402-91 the term “universe” will refer to the “sampling universe,” the group of items which remains after the large dollar or sensitive transactions have been stratified (or segregated) for complete (as opposed to partial) evaluation.
“Variable sampling.” Variable sampling is generally used to verify account balances or cost elements and note any differences. This type of sampling is substantive testing (as opposed to compliance testing) whereby sample items are evaluated for error amounts or variables (as opposed to attributes). The evaluation sampling universe is the entire grouping of items from which a sample will be drawn. Variable sampling can be applied to proposals, incurred costs, and contractor catalog prices.
