People v. Collins
"The People of the State of California v. Collins" was a 1968 jury verdict that famously exploited (and abused) statistics .
Malcolm Ricardo Collins and Janet Louise Collins were convicted of robbing about US $ 35 from a frail lady. They appealed and were acquitted when the California Supreme Court harshly criticized the first instance's deliberations.
Trial in the first instance
Eyewitnesses to a robbery in Los Angeles reported that the perpetrators were a dark-skinned man with a beard and mustache and a light-skinned woman with blond hair tied in a ponytail. They escaped in a yellow car.
After a California college "math instructor" explained the multiplication rule of probability, the prosecutor invited the jury to assess the likelihood that the defendants were not the bank robbers. Although the "instructor" did not address the so-called conditional probability , the prosecutor was certain that the following assumptions were correct:
| Black man with a beard |
1: 10 |
| Man with mustache | 1: 4 |
| White woman with a ponytail |
1: 10 |
| White woman with blond hair |
1: 3 |
| Yellow passenger car | 1: 10 |
| Multiracial couple in a car |
1: 1000 |
The jury found the two defendants guilty. There
They concluded that the accused must inevitably be the guilty party - since the likelihood that there would be another couple in Los Angeles that fulfilled all of the witnesses' descriptions was very small - namely 1 in twelve million. The fact that the victim of the crime could not clearly identify the perpetrators in court and that the information about their clothing contradicted each other played less of a role for the jury than the seemingly impressive numbers.
discussion
From a scientific point of view, the statistical derivation of the expert was criticized because it neglected the existence of conditioned or dependent characteristics - two probabilities may only be multiplied if they are completely independent of one another. For example, a man with a beard is more likely to have a mustache than a beardless man. On the other hand, women are more likely than men to wear ponytails. So when you meet a person with a ponytail, the person is often female. If the probabilities for the cases "ponytail" and "female" were independent, one would have to find the same number of women and men among a random sample of people wearing ponytails.
Another point of discussion in the case was the use of probability theory to determine guilt. Even with a correctly executed probability calculation, there is always the risk of a fallacy. It is also unclear from which degree of probability a guilt can be assumed.