Manufacturers do not contribute to estimates of sampling variability because the manufacturer's mail panel is not a probability sample from a known frame and standard errors of the industry estimates cannot be calculated.
Estimates from all three surveys are also subject to nonsampling errors, which can arise in any stage of the survey. Such errors include coverage error (failure to accurately represent all population units in the sample), response errors, coding errors, and nonresponse.
Although no direct measurement of these errors has been obtained, precautionary steps were taken in all phases of the collection, processing, and tabulation of the data to minimize their influence.