The text with the footnotes is here.
The text without the footnotes is:
[a] Beyond a Reasonable Doubt.
In a criminal tax prosecution, just as any other criminal prosecution, the Government bears the burden of proving guilt beyond a reasonable doubt.
There is no universally accepted jury instruction defining reasonable doubt. Some courts and commentators urge that the bare words – beyond a reasonable doubt – should be used with no attempt to further define the term. Other offerings of instructions provide more words, without perhaps more guidance and clarity. Some perhaps would argue that this is a good state of affairs, for it permits the jury in its collective wisdom to shape the fact finding process to its perception of the needs of the community and the individual charges and defendant.
Trial lawyers love to describe burdens in percentage terms:
Consider what “proof beyond a reasonable doubt” actually mandates that the jury do. Surely it requires more proof than the preponderance of the evidence standard, which governs in civil cases. As commonly explained to civil juries, the preponderance standard is quantified as any amount of certainty greater than 50%, and proof beyond a reasonable doubt must mean more than that. But how much more proof than a preponderance is needed in a criminal case? The quantity of certainty is never quantified; instead, it is kept quite vague. Is 90% certainty required? 95%? 99%? Or could the amount of certainty be much lower, say perhaps 75%?
Indeed, whatever the “percentage” level of certainty imagined to be inherent in the standard, it is reported that “research has consistently shown that the jurors in criminal cases will often be satisfied with much less certainty than is conventionally assumed.”
Judges, when asked to express proof beyond a reasonable doubt as a probability of guilt, generally pick a number between .75 and .90 (depending on the judge), and jury quantifications are similar. These may seem shockingly low figures, implying that as many as a quarter of the people convicted of crime are innocent.
Judge Posner notes that prosecutorial selectivity in picking cases to prosecute substantially mitigates the risk that the innocent will be convicted. Still, assuming that there were no prosecutorial selectivity mitigating factors, are you concerned that judges and jurors in 20 or 25% doubt could return a verdict of guilty in a criminal case?
Judge Posner’s conclusions are his own, based on his experience and his anecdotal polling of his colleagues. Other attempts at empirical studies of these issues show that judges and suggest that jurors are all over the lot on the issue. In one study of his 10 of his colleagues by Judge Weinstein of the Eastern District of New York, “one gave a probability of 76%, one gave 80%, four gave 85%, two gave 90%, and one gave 95%. In other words, the probabilities hovered around 85%-90%.” In a larger survey of federal judges throughout the country:
Of the 171 judges who responded, 126 had thresholds that were 90% or higher. Eleven judges had thresholds of 75% or below, one of whom was satisfied with a 50% probability. The other study was conducted among Illinois state court judges. On a scale of 1 to 10, the mean level of certainty in this study was 8.9, with a median of 8.8; 63% of the judges responded with a level of 9.0 or higher. Most (but not all) judges, then, tend to see the government's burden much the way Blackstone did [about 90%].
There are unique prosecutorial mitigating factors in tax cases that virtually require that the Government only choose cases to prosecute where the evidence in all except the atypical case will produce a conviction. This further mitigates the possibility of conviction of the innocent in tax cases.