Four model variants within a continuous forensic DNA mixture interpretation framework: Effects on evidential...

Area of Science:

• Forensic Science
• Genetics
• Statistical Analysis

Background:

• Probabilistic genotyping (PG) methods for DNA mixture interpretation compute Likelihood Ratios (LRs) but lack standardized models.
• The choice of PG model can influence the LR, affecting evidence weight in legal proceedings.
• Current practices allow individual labs to select models, leading to potential inconsistencies.

Purpose of the Study:

• To investigate the impact of different continuous mixture interpretation models on Likelihood Ratio (LR) calculations.
• To assess the variability in LR results and their verbal expressions across distinct PG models.
• To quantify the influence of model choice on forensic evidence interpretation.

Main Methods:

• Four variants of a continuous mixture interpretation method were employed.
• Models were tested on 101 experimental samples with 1, 2, and 3 known contributors.
• LRs were computed for each sample, comparing results across the four models.

Main Results:

• Intra-model variability increased with more contributors and lower contributor DNA template mass.
• Inter-model variability in verbal LR expression was observed in 32 of 195 comparisons.
• In 11 cases, the LR changed from >1 to <1 between models, altering interpretation.

Conclusions:

• Differences in PG models can significantly impact the calculated LR and its verbal categorization.
• Variability underscores the need for standardized guidelines and further research into model reliability.
• Independent studies are crucial to quantify the reliability and variability of forensic DNA interpretation systems.