Forensic Evidence Interpretation and Statistics

To evaluate the strength and limitations of evidence using relevant probabilities, data, and clearly defined propositions.

A specific statement about an event, source, or activity whose support can be evaluated using forensic evidence.

The probability of observing the evidence under one proposition divided by its probability under an alternative proposition.

LR = P(Evidence given H1) divided by P(Evidence given H2), where H1 and H2 are competing propositions.

The observed evidence is more probable under the proposition in the numerator than under the alternative proposition.

The observed evidence is more probable under the alternative proposition in the denominator.

The evidence provides no relative support for either proposition under the specified model.

4, indicating the evidence is four times as probable under H1 as under H2.

0.2, indicating the evidence is five times as probable under H2 as under H1.

The probability of an event given that another specified event or condition is known.

To update the probability of a proposition in light of evidence using prior probabilities and the likelihood of that evidence.

Posterior odds = prior odds × likelihood ratio.

The relative probabilities of competing propositions before considering the particular evidence being evaluated.

The relative probabilities of competing propositions after incorporating the evidence through an appropriate likelihood ratio.

2:1 in favor of H1.

90%, assuming the propositions are mutually exclusive and exhaustive and the model is valid.

Confusing the probability of evidence given innocence or an alternative proposition with the probability of innocence given the evidence.

Improperly treating a possible coincidental match as insignificant solely because many potential matching individuals exist.

Mistakenly treating P(A given B) as equivalent to P(B given A).

A random match probability concerns the chance of a matching profile under specified conditions, not the probability of guilt or innocence.

The estimated probability that an unrelated individual from a specified population would have a matching DNA profile under the model used.

The proportion of copies of a particular genetic variant among the relevant alleles in a population.

The proportion or modeled probability of a particular combination of alleles at a genetic locus in a population.

A population-genetic model under which genotype frequencies can be predicted from allele frequencies under specified assumptions.

2pq.

p squared.

0.12, calculated as 2 × 0.2 × 0.3.

0.01, calculated as 0.1 squared.

Multiplication of appropriate genotype frequencies across loci when the model's independence assumptions and adjustments are satisfied.

Genetic variation among subpopulations that can affect genotype frequency estimates and require appropriate modeling.

A population-substructure adjustment used in certain genotype frequency calculations.

A proposition concerning the origin of a material or trace, such as whether a particular person contributed DNA.

A proposition concerning how or through what activity evidence was deposited or transferred.

A proposition concerning whether a criminal act occurred or a particular person committed it.

Identifying a possible source does not by itself determine when, why, or how the material was transferred.

DNA from one person can reach an object through another person or surface without direct contact by the original contributor.

The alternative proposition used to calculate the probability of observing the evidence for comparison with the numerator proposition.

The meaning and strength of a likelihood ratio depend on the precise propositions being compared.

The population or reference group appropriate to the specific propositions, circumstances, and statistical model.

A systematic distortion caused when collected samples do not adequately represent the population of interest.

An interval produced by a statistical procedure designed to contain the target parameter at a stated long-run coverage rate.

The possibility that a reported positive association is incorrect, particularly when testing procedures and decision thresholds are not properly considered.

A predefined criterion used to classify or interpret analytical results under a specified method.

A precise numerical statement requires suitable data, validated methods, and assumptions that justify the claimed precision.

Evidential strength concerns how strongly findings support specified propositions; legal conclusions require assessment of all evidence under applicable law.

Strong support for a source-level association versus the separate question of how and when the DNA was deposited.

The prosecutor's fallacy, involving an improper reversal of conditional probabilities.

The observed findings are 20 times as probable under the specified common-source proposition as under the specified alternative, assuming the model is appropriate.

No. Likelihood ratios are proposition-dependent; both calculations must be evaluated for relevance, assumptions, and validity.

No. The likelihood ratio expresses relative support for specified propositions and does not independently determine guilt or the legal standard of proof.