|
Some of the 2000 data in this Appendix have been estimated using standard statistical techniques. To give the reader a better idea
of the accuracy of these estimates, they are reported along with the "relative standard error" (RSE). Ordinarily, it is expected that
on approximately two out of three occasions, the true value of a statistic, such as a total, will be within one standard error, either
high or low, of the estimated statistic. A relative standard error is the standard error divided by the statistic (for example, a total)
expressed as a percent. The smaller the RSE, the smaller is the range around the estimate and, therefore, the more confidence one
can have in using the estimate. (Another name for relative standard error is "coefficient of variation," or CV, and the EIA has often
used that term.)
For example, if a total is estimated to be 213.3, with an estimated RSE of 5 percent, then there are about two chances in three that
the actual total is between 202.6 and 224.0. Nonsampling error is not fully taken into account, and may make results substantially
less accurate. Thus, a more reasonable estimated total to report, along with an estimated RSE of 5 percent, would be 210.
Technical information regarding the sources and quality of the data in this report is available in the Technical Notes of the Electric
Power Monthly, DOE/EIA-0226. That report is accessible via the Internet at: http://www.eia.doe.gov/cneaf/electricity/epm/epm_sum.html.
|