Accuracy and reliability of forensic handwriting comparisons.

Forensic handwriting examination involves the comparison of writing samples by forensic document examiners (FDEs) to determine whether or not they were written by the same person. Here we report the results of a large-scale study conducted to assess the accuracy and reliability of handwriting comparison conclusions. Eighty-six practicing FDEs each conducted up to 100 handwriting comparisons, resulting in 7,196 conclusions on 180 distinct comparison sets, using a five-level conclusion scale. Erroneous "written by" conclusions (false positives) were reached in 3.1% of the nonmated comparisons, while 1.1% of the mated comparisons yielded erroneous "not written by" conclusions (false negatives). False positive rates were markedly higher for nonmated samples written by twins (8.7%) compared to nontwins (2.5%). Notable associations between training and performance were observed: FDEs with less than 2 y of formal training generally had higher error rates, but they also had higher true positive and true negative rates because they tended to provide more definitive conclusions; FDEs with at least 2 y of formal training were less likely to make definitive conclusions, but those definitive conclusions they made were more likely to be correct (higher positive predictive and negative predictive values). We did not observe any association between writing style (cursive vs. printing) and rates of errors or incorrect conclusions. This report also provides details on the repeatability and reproducibility of conclusions, and reports how conclusions are affected by the quantity of writing and the similarity of content.

Accuracy, reproducibility, and repeatability of forensic footwear examiner decisions.

The interpretation of footwear evidence relies on the expertise of forensic footwear examiners. Here we report on the largest study to date of the accuracy, reproducibility (inter-examiner variation), and repeatability (intra-examiner variation) of footwear examiners' decisions. In this study, 84 practicing footwear examiners each conducted up to 100 comparisons between questioned footwear impressions (provided as photographs and digital images) and known footwear (provided as photographs, transparent test impressions, and digital images), resulting in a total of 6610 comparisons. The quality and characteristics of the impressions were selected to be broadly representative of those encountered in casework. A multilevel conclusion scale was used: 40% of responses were definitive conclusions (identification or exclusion), 14% probable conclusions (high degree of association or indications of non-association), 40% class associations (association of class characteristics or limited association of class characteristics), and 6% neutral conclusions (inconclusive or not suitable). On nonmated comparisons, 0.2% of conclusions were erroneous identifications (false positives), and 1.4% were incorrect responses of "high degree of association." The majority of erroneous identifications were made by a single participant. On mated comparisons, 6.0% of conclusions were erroneous exclusions (false negatives), and 1.8% were incorrect responses of "indications of non-association." Erroneous conclusions were sometimes reproduced by different examiners, but rarely repeated by the same examiner-1.1% of erroneous identifications were reproduced (none were repeated) and 19.9% of erroneous exclusions were reproduced (just one was repeated). Examiners' assessments of whether a questioned impression was suitable for comparison were notably inconsistent and may benefit from standardization. Rates of correct definitive conclusions are directly associated with the quality of the questioned impression and the extent of class similarities/differences between the questioned impression and known footwear.

Elucidating the relationships between two automated handwriting feature quantification systems for multiple pairwise comparisons.

Recent advances in complex automated handwriting identification systems have led to a lack of understandability of these systems' computational processes and features by the forensic handwriting examiners that they are designed to support. To mitigate this issue, this research studied the relationship between two systems: FLASH ID® , an automated handwriting/black box system that uses measurements extracted from a static image of handwriting, and MovAlyzeR® , a system that captures kinematic features from pen strokes. For this study, 33 writers each wrote 60 phrases from the London Letter using cursive writing and handprinting, which led to thousands of sample pairs for analysis. The dissimilarities between pairs of samples were calculated using two score functions (one for each system). The observed results indicate that dissimilarity scores based on kinematic spatial-geometric pen stroke features (e.g., amplitude and slant) have a statistically significant relationship with dissimilarity scores obtained using static, graph-based features used by the FLASH ID® system. Similar relationships were observed for temporal features (e.g., duration and velocity) but not pen pressure, and for both handprinting and cursive samples. These results strongly imply that both the current implementation of FLASH ID® and MovAlyzeR® rely on similar features sets when measuring differences in pairs of handwritten samples. These results suggest that studies of biometric discrimination using MovAlyzeR® , specifically those based on the spatial-geometric feature set, support the validity of biometric matching algorithms based on FLASH ID® output.

Characterization and differentiation of aluminum powders used in improvised explosive devices. Part 2: Micromorphometric method refinement and preliminary statistical analysis.

Aluminum (Al) powder is commonly encountered in improvised explosive devices (IEDs) as a metallic fuel due to its availability and low cost. Although available commercially in powder form, amateur bomb-makers also produce their own Al powder via simple methods found online. In order to provide investigative leads and forensic intelligence, it is important to evaluate not only the composition of homemade devices, but also to distinguish between the various forms of Al powder they contain. To achieve this goal, a method using automated microscopy in combination with statistical techniques has been demonstrated to have the potential to provide source discrimination and investigative leads in source attribution of Al powders in IEDs. The present research refined this method and investigated 59 industrially and amateurly produced Al powder sources with seven subsamples per source using two traditional linear discriminant analyses (LDA), one with a standard data split for training and testing, and another using leave-one-out cross-validation. Averaging the classification accuracies for the two LDA-based analyses, LDA has the ability to correctly classify 59.26%, 83.35%, and 80.69% of the samples based on their powder source, type, and production method, respectively. This classification accuracy represents a 3407%, 317%, and 61.38% increase in accuracy from random class assignment, respectively. Further, in most instances of incorrect data attribution to a particular source, the subsample has been misidentified with another sample of the same powder type or production method.

Characterizing missed identifications and errors in latent fingerprint comparisons using eye-tracking data.

Latent fingerprint examiners sometimes come to different conclusions when comparing fingerprints, and eye-gaze behavior may help explain these outcomes. missed identifications (missed IDs) are inconclusive, exclusion, or No Value determinations reached when the consensus of other examiners is an identification. To determine the relation between examiner behavior and missed IDs, we collected eye-gaze data from 121 latent print examiners as they completed a total 1444 difficult (latent-exemplar) comparisons. We extracted metrics from the gaze data that serve as proxies for underlying perceptual and cognitive capacities. We used these metrics to characterize potential mechanisms of missed IDs: Cursory Comparison and Mislocalization. We find that missed IDs are associated with shorter comparison times, fewer regions visited, and fewer attempted correspondences between the compared images. Latent print comparisons resulting in erroneous exclusions (a subset of missed IDs) are also more likely to have fixations in different regions and less accurate correspondence attempts than those comparisons resulting in identifications. We also use our derived metrics to describe one atypical examiner who made six erroneous identifications, four of which were on comparisons intended to be straightforward exclusions. The present work helps identify the degree to which missed IDs can be explained using eye-gaze behavior, and the extent to which missed IDs depend on cognitive and decision-making factors outside the domain of eye-tracking methodologies.

Assessing the microscale heterogeneity in Standard Reference Material 4600 Surrogate Post-detonation Urban Debris.

Nondestructive microbeam X-ray fluorescence (µXRF) spectrometry has been used to investigate the elemental microheterogeneity in a nuclear forensics reference material (RM), NIST SRM 4600 Surrogate Post-detonation Urban Debris. Using a principal component analysis (PCA) model, results indicate the majority of elements appear homogeneous; however, zinc (Zn) exhibits microscale heterogeneity for this SRM. To minimize contributions to the measurement uncertainty from elemental microheterogeneity, a minimum sample mass of 24 mg is recommended for analysis.

Characterization and differentiation of aluminum powders used in improvised explosive devices - Part 1: Proof of concept of the utility of particle micromorphometry.

Aluminum (Al) powders are commonly used in improvised explosive devices as metallic fuels, a component of explosive mixtures. These powders can be obtained readily from industrial-scale and consumer products, and produced using unsophisticated "kitchen chemistry" techniques. This research demonstrates the potential of automated particle micromorphometry for comparisons between known source and questioned Al powders recovered from IEDs, as well as for insight into the method of Al powder manufacture. Al powder samples were obtained from legitimate manufacturers, and 56 samples were produced "in-house" from Al-containing spray paints and ball-milled Al foils. Transmitted light microscope images of Al powder particles were acquired using an automated stage with automated z-focus; 17 size and shape parameters were measured for all particles. Approximately 37,000-2,500,000 particles/sample were analyzed using an open-source statistical package with customized code. Dimensionality reduction was required for processing the large datasets: eight of the 17 measured variables were selected based on inspection of the correlation matrix. Data from four subsamples from each of the 56 samples produced using "in-house" methods were analyzed using ANOVA to assess the within- and between-sample variation. High within-sample variation was noted; however, ANOVA and post-hoc Tukey's honestly significant difference (HSD) tests demonstrated that the between-sample variation was substantially larger than the within-sample variation. Each sample could be differentiated from all other samples in the test set. Future experiments will focus on ways to reduce the within-sample variation, and additional statistical and microanalytical methods to classify sources and confidently constrain the method of Al powder manufacture.

Why do latent fingerprint examiners differ in their conclusions?

Forensic latent print examiners usually but do not always reproduce each other's conclusions. Using data from tests of experts conducting fingerprint comparisons, we show the extent to which differing conclusions can be explained in terms of the images, and in terms of the examiners. Some images are particularly prone to disagreements or erroneous conclusions; the highest and lowest quality images generally result in unanimous conclusions. The variability among examiners can be seen as the effect of implicit individual decision thresholds, which we demonstrate are measurable and differ substantially among examiners; this variation may reflect differences in skill, risk tolerance, or bias. Much of the remaining variability relates to inconsistency of the examiners themselves: borderline conclusions (i.e., close to individual decision thresholds) often were not repeated by the examiners themselves, and tended to be completed more slowly and rated difficult. A few examiners have significantly higher error rates than most: aggregate error rates of many examiners are not necessarily representative of individual examiners. The use of a three-level conclusion scale does not precisely represent the underlying agreements and disagreements among examiners. We propose a new method of quantifying examiner skill that would be appropriate for use in proficiency tests. These findings are operationally relevant to staffing, quality assurance, and disagreements among experts in court.

Gaze behavior and cognitive states during fingerprint target group localization.

BACKGROUND: The comparison of fingerprints by expert latent print examiners generally involves repeating a process in which the examiner selects a small area of distinctive features in one print (a target group), and searches for it in the other print. In order to isolate this key element of fingerprint comparison, we use eye-tracking data to describe the behavior of latent fingerprint examiners on a narrowly defined "find the target" task. Participants were shown a fingerprint image with a target group indicated and asked to find the corresponding area of ridge detail in a second impression of the same finger and state when they found the target location. Target groups were presented on latent and plain exemplar fingerprint images, and as small areas cropped from the plain exemplars, to assess how image quality and the lack of surrounding visual context affected task performance and eye behavior. One hundred and seventeen participants completed a total of 675 trials. RESULTS: The presence or absence of context notably affected the areas viewed and time spent in comparison; differences between latent and plain exemplar tasks were much less significant. In virtually all trials, examiners repeatedly looked back and forth between the images, suggesting constraints on the capacity of visual working memory. On most trials where context was provided, examiners looked immediately at the corresponding location: with context, median time to find the corresponding location was less than 0.3 s (second fixation); however, without context, median time was 1.9 s (five fixations). A few trials resulted in errors in which the examiner did not find the correct target location. Basic gaze measures of overt behaviors, such as speed, areas visited, and back-and-forth behavior, were used in conjunction with the known target area to infer the underlying cognitive state of the examiner. CONCLUSIONS: Visual context has a significant effect on the eye behavior of latent print examiners. Localization errors suggest how errors may occur in real comparisons: examiners sometimes compare an incorrect but similar target group and do not continue to search for a better candidate target group. The analytic methods and predictive models developed here can be used to describe the more complex behavior involved in actual fingerprint comparisons.

Timing and Appearance of Postmortem Root Banding in Nonhuman Mammals.

A study was undertaken using nonhuman mammal specimens to better understand environmental influences on postmortem hair root band (PMRB) formation and to see whether PMRBs would occur in nonhuman mammal hairs in a similar fashion to human hairs. Carcasses from surrounding roadways were the primary source of specimens for this study, augmented by donated deceased domestic pets. Sections of pelt from each specimen were placed in controlled environmental conditions while the remainder of the carcass was left in a secure outdoor setting. Hair samples were collected daily from outdoor and control specimens and examined for evidence of PMRBs. Several environmental factors were also recorded on a daily basis. Results demonstrate PMRBs can occur in nonhuman mammal hairs, and they have microscopic characteristics similar to human PMRBs. Factors found to correlate with PMRB formation include postmortem interval, temperature, pH, and the formation and subsequent volatilization of ammonia from the surrounding tissue.

Factors associated with latent fingerprint exclusion determinations.

Exclusion is the determination by a latent print examiner that two friction ridge impressions did not originate from the same source. The concept and terminology of exclusion vary among agencies. Much of the literature on latent print examination focuses on individualization, and much less attention has been paid to exclusion. This experimental study assesses the associations between a variety of factors and exclusion determinations. Although erroneous exclusions are more likely to occur on some images and for some examiners, they were widely distributed among images and examiners. Measurable factors found to be associated with exclusion rates include the quality of the latent, value determinations, analysis minutia count, comparison difficulty, and the presence of cores or deltas. An understanding of these associations will help explain the circumstances under which errors are more likely to occur and when determinations are less likely to be reproduced by other examiners; the results should also lead to improved effectiveness and efficiency of training and casework quality assurance. This research is intended to assist examiners in improving the examination process and provide information to the broader community regarding the accuracy, reliability, and implications of exclusion decisions.

A Set of Handwriting Features for Use in Automated Writer Identification.

A writer's biometric identity can be characterized through the distribution of physical feature measurements ("writer's profile"); a graph-based system that facilitates the quantification of these features is described. To accomplish this quantification, handwriting is segmented into basic graphical forms ("graphemes"), which are "skeletonized" to yield the graphical topology of the handwritten segment. The graph-based matching algorithm compares the graphemes first by their graphical topology and then by their geometric features. Graphs derived from known writers can be compared against graphs extracted from unknown writings. The process is computationally intensive and relies heavily upon statistical pattern recognition algorithms. This article focuses on the quantification of these physical features and the construction of the associated pattern recognition methods for using the features to discriminate among writers. The graph-based system described in this article has been implemented in a highly accurate and approximately language-independent biometric recognition system of writers of cursive documents.

Microscopical characterization of known postmortem root bands using light and scanning electron microscopy.

A postmortem root band (PMRB) is a distinct microscopic feature that is postulated to occur in hair remaining in the follicle during the postmortem interval [1] (Petraco et al., 1998). The scientific validity of this premise has been highlighted in two recent high-profile criminal cases involving PMRBs [2,3] (State of Florida v. Casey Marie Anthony, 2008; People v. Kogut, 2005). To better understand the fundamental aspects of postmortem root banding, the microscopical properties of known PMRBs1 were characterized by light microscopy, and scanning electron microscope (SEM) imaging of microtomed sections of hairs showing root banding. The results from this study show that the appearance of the PMRB may be due to the degradation of the chemically labile, non-keratin intermacrofibrillar matrix (IMM) in the pre-keratin/keratogenous region of anagen hairs. In addition, this degradation is confined to the cortex of the hair, with no apparent damage to the layers of the cuticle. These results could provide valuable information for determining the mechanism of band formation, as well as identify a set of microscopic features that could be used to distinguish hairs with known PMRBs from similarly looking environmentally degraded hairs.

Data on the interexaminer variation of minutia markup on latent fingerprints.

The data in this article supports the research paper entitled "Interexaminer variation of minutia markup on latent fingerprints" [1]. The data in this article describes the variability in minutia markup during both analysis of the latents and comparison between latents and exemplars. The data was collected in the "White Box Latent Print Examiner Study," in which each of 170 volunteer latent print examiners provided detailed markup documenting their examinations of latent-exemplar pairs of prints randomly assigned from a pool of 320 pairs. Each examiner examined 22 latent-exemplar pairs; an average of 12 examiners marked each latent.

Interexaminer variation of minutia markup on latent fingerprints.

Latent print examiners often differ in the number of minutiae they mark during analysis of a latent, and also during comparison of a latent with an exemplar. Differences in minutia counts understate interexaminer variability: examiners' markups may have similar minutia counts but differ greatly in which specific minutiae were marked. We assessed variability in minutia markup among 170 volunteer latent print examiners. Each provided detailed markup documenting their examinations of 22 latent-exemplar pairs of prints randomly assigned from a pool of 320 pairs. An average of 12 examiners marked each latent. The primary factors associated with minutia reproducibility were clarity, which regions of the prints examiners chose to mark, and agreement on value or comparison determinations. In clear areas (where the examiner was "certain of the location, presence, and absence of all minutiae"), median reproducibility was 82%; in unclear areas, median reproducibility was 46%. Differing interpretations regarding which regions should be marked (e.g., when there is ambiguity in the continuity of a print) contributed to variability in minutia markup: especially in unclear areas, marked minutiae were often far from the nearest minutia marked by a majority of examiners. Low reproducibility was also associated with differences in value or comparison determinations. Lack of standardization in minutia markup and unfamiliarity with test procedures presumably contribute to the variability we observed. We have identified factors accounting for interexaminer variability; implementing standards for detailed markup as part of documentation and focusing future training efforts on these factors may help to facilitate transparency and reduce subjectivity in the examination process.

Changes in latent fingerprint examiners' markup between analysis and comparison.

After the initial analysis of a latent print, an examiner will sometimes revise the assessment during comparison with an exemplar. Changes between analysis and comparison may indicate that the initial analysis of the latent was inadequate, or that confirmation bias may have affected the comparison. 170 volunteer latent print examiners, each randomly assigned 22 pairs of prints from a pool of 320 total pairs, provided detailed markup documenting their interpretations of the prints and the bases for their comparison conclusions. We describe changes in value assessments and markup of features and clarity. When examiners individualized, they almost always added or deleted minutiae (90.3% of individualizations); every examiner revised at least some markups. For inconclusive and exclusion determinations, changes were less common, and features were added more frequently when the image pair was mated (same source). Even when individualizations were based on eight or fewer corresponding minutiae, in most cases some of those minutiae had been added during comparison. One erroneous individualization was observed: the markup changes were notably extreme, and almost all of the corresponding minutiae had been added during comparison. Latents assessed to be of value for exclusion only (VEO) during analysis were often individualized when compared to a mated exemplar (26%); in our previous work, where examiners were not required to provide markup of features, VEO individualizations were much less common (1.8%).

A novel approach for latent print identification using accurate overlays to prioritize reference prints.

A novel approach to automated fingerprint matching and scoring that produces accurate locally and nonlinearly adjusted overlays of a latent print onto each reference print in a corpus is described. The technology, which addresses challenges inherent to latent prints, provides the latent print examiner with a prioritized ranking of candidate reference prints based on the overlays of the latent onto each candidate print. In addition to supporting current latent print comparison practices, this approach can make it possible to return a greater number of AFIS candidate prints because the ranked overlays provide a substantial starting point for latent-to-reference print comparison. To provide the image information required to create an accurate overlay of a latent print onto a reference print, "Ridge-Specific Markers" (RSMs), which correspond to short continuous segments of a ridge or furrow, are introduced. RSMs are reliably associated with any specific local section of a ridge or a furrow using the geometric information available from the image. Latent prints are commonly fragmentary, with reduced clarity and limited minutiae (i.e., ridge endings and bifurcations). Even in the absence of traditional minutiae, latent prints contain very important information in their ridges that permit automated matching using RSMs. No print orientation or information beyond the RSMs is required to generate the overlays. This automated process is applied to the 88 good quality latent prints in the NIST Special Database (SD) 27. Nonlinear overlays of each latent were produced onto all of the 88 reference prints in the NIST SD27. With fully automated processing, the true mate reference prints were ranked in the first candidate position for 80.7% of the latents tested, and 89.8% of the true mate reference prints ranked in the top ten positions. After manual post-processing of those latents for which the true mate reference print was not ranked first, these frequencies increased to 90.9% (1st rank) and 96.6% (top ten), respectively. Because the computational process is highly parallelizable, it is feasible for this method to work with a reference corpus of several thousand prints.

Measuring what latent fingerprint examiners consider sufficient information for individualization determinations.

Latent print examiners use their expertise to determine whether the information present in a comparison of two fingerprints (or palmprints) is sufficient to conclude that the prints were from the same source (individualization). When fingerprint evidence is presented in court, it is the examiner's determination--not an objective metric--that is presented. This study was designed to ascertain the factors that explain examiners' determinations of sufficiency for individualization. Volunteer latent print examiners (n = 170) were each assigned 22 pairs of latent and exemplar prints for examination, and annotated features, correspondence of features, and clarity. The 320 image pairs were selected specifically to control clarity and quantity of features. The predominant factor differentiating annotations associated with individualization and inconclusive determinations is the count of corresponding minutiae; other factors such as clarity provided minimal additional discriminative value. Examiners' counts of corresponding minutiae were strongly associated with their own determinations; however, due to substantial variation of both annotations and determinations among examiners, one examiner's annotation and determination on a given comparison is a relatively weak predictor of whether another examiner would individualize. The extensive variability in annotations also means that we must treat any individual examiner's minutia counts as interpretations of the (unknowable) information content of the prints: saying "the prints had N corresponding minutiae marked" is not the same as "the prints had N corresponding minutiae." More consistency in annotations, which could be achieved through standardization and training, should lead to process improvements and provide greater transparency in casework.