Lately, U.S. Presidential candidates George W. Bush and Albert Gore Jr. have been bombarding the public with made to order, fast food statistics to satiate their hunger for a political grist. Like fast food, the statistics they have been dishing out have very little value.

“Almost half of Texas prisoners go back behind bars in three years” was one statistic served up by Gore during a recent speech, the inferences being: (1) Bush is responsible, and (2) 50% is an unacceptable rate of recidivism. His statistic, however, provides little support for the conclusions he would like his audience to draw. First, to conclude that Bush is responsible, the observed data must be relevant to the timing of Bush’s policy, taking into account possible time lag for the policy to have an impact. In fact, due to the long time lag in analyzing how many convicts return to prison after three years, figures have not yet been available for inmates released since Bush became governor in 1995. Consequently, in the absence of any time reference for the observed data, it is impossible to conclude who is responsible for the statistic. Second, without a benchmark with which to compare, it is difficult to make any judgment about a 50% recidivism rate, good or bad. If 50% is below the national average or it is an improvement from the prior period, then it might well be an acceptable rate.

On the other hand, “federal prisons have cut drug-treatment beds by one-third” was Bush’s statistic for criticizing Clinton-Gore’s federal prison drug-treatment policy. The inference here is that Clinton-Gore administration has reduced support for drug-treatment programs for federal prisoners. Although the statistic is supported by figures from the Justice Department, the inference is not. To begin with, you cannot conclude that reduction in spending on treatment beds translates to an overall reduction in spending on drug-treatment programs. In fact, according to a recent AP article, "while the percentage of prisoners in mainline drug treatment dropped by about a third, the percentage in alternative forms of drug counseling doubled from 1991 to 1997.” Moreover, money may not be a measure of efficacy. The spending cut may be a result of improvements in effectiveness and efficiency of drug-treatment programs.

For right or wrong, statistics are a powerful tool for credibility and persuasion, and no one seems to understand this any better than politicians.

More Lies, Damned Lies