As discussed in previous posts, forensic alcohol analysis (the measurement of blood alcohol levels for use as evidence) is based upon the premise that the subject is an "average person" — a premise that simply does not exist in the vast majority of cases. An example of this is the individual who has a higher percentage of body fat than normal. Alcohol is distributed throughout the body according to the water content of blood and tissues. It is not, however, soluble in fat. Thus, if the tested subject has a high percentage of fatty tissue in his body, there will be a lower "volume of distribution" — that is, a smaller percentage of body mass absorbing the alcohol. See Hawkins and Kalant, "The Metabolism of Ethanol and Its Metabolic Effects", 24 Pharmacological Review 67 (1972).
So what does all of this mean in a DUI case? Well, it is not illegal to have a blood alcohol concentration (BAC) of .08% or higher at the police station where the breath test takes place — only at the time of driving. Therefore, it often becomes necessary to estimate the earlier BAC based upon known factors (weight, sex) and assumed factors (rates of absorption and elimination of alcohol in the fictional "average person"); this process is called retrograde extrapolation.
The formula used for this process of estimating blood alcohol levels in the "average person" is called the Widmark factor. However, although the formula takes into account body weight, it does not consider what percentage of that weight is fatty tissue; it simply assumes that it falls within a relatively normal range (called a "body mass index"). Since the formula varies inversely with the volume of distribution, the elimination of alcohol from the body will be quicker in a fat person — thus further skewing an already inaccurate attempt at guessing what the suspect’s BAC was when at the wheel. Woman, parenthetically, will usually experience a faster rate of elimination since the "average woman" has a higher percentage of body fat than the "average man".