Manufacturing Energy Consumption Survey

The sample is updated before each survey to account for business births and deaths since the prior sample.

The establishments included in the 1998 and 2002 MECS were selected using a modified version of the Tillé sampling procedures. The basic idea of Tillé sampling is to start with all of the establishments in the universe for a particular stratum, and discard establishments one-by-one until the desired sample size is left. Before this could be done, each establishment in the stratum universe needed to be assigned a probability of being included in the sample. These probabilities were derived with a probability-proportionate-to-size (PPS) approach. Thus, the inclusion probabilities are proportionate to their respective MOS, which was cost of energy.

Once these probabilities were set, the rejection procedure could be carried out. The procedure is an iterative process. It begins with all of the establishments in the industry-by-region stratum eligible for the sample. At each subsequent stage, a randomly selected establishment is thrown out from consideration. Establishments with very large MOS relative to the other establishments in the stratum have a better of chance of not getting thrown out at a particular stage than those with relatively smaller MOS. If the number of establishments in the universe is N and the desired sample size is n, then (N-n) iterations are required.

The Manufacturing Energy Consumption Survey (MECS) is a mail-out/mail-back survey of establishments selected from the Census of Manufactures or Annual Survey of Manufactures. Establishments for the most energy intensive industries are selected. Establishments in other industries are stratified by industry and geographic location, and selected with probabilities proportional to total energy costs. Selected establishments receive one of three industry-based forms--one for petroleum refiners; one for primary chemical, paper, primary metal, and wood product producers; and one for all other manufacturers. The survey was conducted every 3 years from 1985 through 1991, and quadrennially beginning in 1994. For more information on survey methodology, see https://www.eia.gov/consumption/manufacturing/data/2002/index.php?view=methodology

To link to external business data, also request County Business Patterns Business Register.

• Employer Identification Number (EIN)
• Census region
• State FIPS code
• Zip code
• Other (see Linkage Capabilities description)