Inter-Laboratory Variation in Conclusions and Statistical Analyses for DNA Mixtures

The purpose of the presentation is to discuss the results collected and variation observed in interpretations, conclusions, and statistical analyses reported by forensic laboratories when assessing DNA mixture profiles with respect to person of interest (POI) DNA profiles. These results were collected as part of the NIJ-funded (# 2020-R2-CX- 0049) study entitled Inter-laboratory Variation in Interpretation of DNA Mixtures (hereafter referred to as “DNAmix 2021”).

DNAmix 2021 is a rigorous, large-scale study conducted to evaluate the extent of consistency and variation among forensic laboratories in interpretations and statistical analyses of DNA mixtures, and to assess the effects of various potential sources of variability. To this end, this study utilized a multi-phased approach. This presentation reports results from the final phase — the Interpretation, Comparison, and Statistical Analysis (ICSA) Subtest. The earlier phases addressed laboratory policies and procedures relevant to DNA mixture interpretation, analysis procedures or decisions that may vary depending upon the case scenario, the nature of mixture casework conducted by laboratories, and assessments of suitability and number of contributors.

Analyses are based on 765 responses reported by 106 participants representing 52 forensic laboratories. The ICSA Subtest was based on eight distinct DNA mixtures, provided to participants as electropherograms. Each DNA mixture was provided to participants with a POI reference profile either as a contributor packet (in which the POI was in the mixture), or as a noncontributor packet (in which the POI was NOT in the mixture). These mixtures were created under controlled conditions to allow for ground truth attribution, and were designed and created to be broadly representative of the range of attributes encountered in casework. All DNA profiles used in this study (mixtures and references) were from real people; no profiles were simulated or contrived.

During the ICSA Subtest, participants submitted responses regarding their comparisons (the categorical conclusion that would be reported with respect to the POI) and their statistical analyses with respect to the POI (the statistical values computed using each population database routinely used in operational casework). During this presentation, we will report on the accuracy and inter- laboratory variability of categorical conclusions, the accuracy and variability of statistical responses, and the degree to which sample-specific factors or laboratory standard operating procedures explain any observed variation.