A series of nine working papers, chronologically sequenced, are the result of a study to develop a statistical methodology and to formulate specific statistical procedures for the analysis of in-use automobile fuel-economy data. The methodology developed addresses specific aspects of the following three issues: difference between actual, in-use gas mileage numbers and those reported in EPA (Environmental Protection Agency) certification tests; the reliability of EPA fuel-economy figures in providing a relative measure of gas mileage; and the tendancy for the discrepancy between EPA and in-use gas mileage figures to increase from one model year to another. To a considerable extent, however, the methodology represents adaptation of statistical techniques to generic questions which are not unique to the fuel-economy problem and can, accordingly, be construed in a much broader context. The topics covered by the working papers include upward bias of variance computed relative to the harmonic mean; bases for statistical estimation of fuel economy; estimating the standard error for a sample harmonic mean; considerations pertaining to the effect of accumulated odometer miles on mpg shortfalls; aggregation of data from the sources with groups of "replicate" vehicles; philosophical and analytical issues affecting the federal energy administration fuel-economy study; a regression/covariance analysis approach to estimation of fuel economy; some considerations affecting vehicle ranking according to fuel economy; and a regression approach to that ranking.

  • Corporate Authors:

    Falcon Research and Development

    1 American Drive
    Buffalo, NY  United States  14225
  • Publication Date: 0

Subject/Index Terms

Filing Info

  • Accession Number: 00321346
  • Record Type: Publication
  • Source Agency: Transportation Systems Center
  • Files: TSR
  • Created Date: Oct 30 1982 12:00AM