## Facts

- Juanita Brooks was robbed in an alley in the San Pedro area of Los Angeles on June 18, 1964; she was pushed down, saw a young woman fleeing, and her purse containing about $35–$40 was taken.
- A witness, John Bass, saw a woman run from the alley and enter a yellow car driven by a Black man.
- Malcolm Ricardo Collins (a Black man) and Janet Louise Collins (a white woman) were prosecuted as the perpetrators and convicted by a jury of second-degree robbery.
- The prosecution introduced eyewitness and descriptive evidence and, to strengthen identification, called a mathematics instructor to present probability calculations based on a set of characteristics attributed to the perpetrators (including a yellow car, a Black man with facial hair, a white woman with blond hair in a ponytail, and an interracial couple).
- The prosecution assigned numerical probabilities to individual traits without empirical support and multiplied them to claim that only “1 in 12 million” couples would share all the listed characteristics.
- The prosecutor argued from this calculation that it was extraordinarily unlikely that anyone other than the defendants committed the robbery.
- Only Malcolm Collins appealed, challenging the admissibility and use of the probability evidence.

## Issues

1. Whether mathematical probability testimony was admissible when the numerical estimates lacked empirical support and accepted statistical methodology.
2. Whether the multiplication of trait probabilities improperly assumed statistical independence without proof.
3. Whether the prosecution’s probabilistic presentation and argument misled the jury and risked diluting the requirement of proof beyond a reasonable doubt.

## Decision

- The California Supreme Court reversed Malcolm Collins’s conviction and remanded for a new trial.
- The court held the probability evidence, as presented and argued, lacked an adequate foundation and was used in a misleading manner.
- The court concluded that the probability testimony created a substantial risk of distorting the jury’s fact-finding role and the reasonable-doubt standard.

## Legal Principles

- Mathematical or statistical evidence in criminal trials is not categorically barred, but it must satisfy ordinary evidentiary requirements, including adequate foundational support.
- Probability estimates must be grounded in reliable empirical data and accepted methodology; unsupported numerical assumptions are improper.
- The multiplication rule may not be used to generate a composite probability unless any assumption of independence among traits is justified.
- A low probability that a random person would match a description cannot be treated as a correspondingly high probability of guilt; arguing otherwise risks misleading the jury.
- Courts may exclude or reverse where numeric evidence carries an undue risk of confusing or overawing jurors and thereby undermining proof beyond a reasonable doubt.

## Conclusion

The court ordered a new trial because the prosecution’s “one in 12 million” probability presentation rested on speculative inputs and unjustified assumptions, and it was argued in a way that could mislead jurors into treating numerical odds as a substitute for proof beyond a reasonable doubt.


People v. Collins, 68 Cal. 2d 319 (Cal. 1968)