Annabel Bolck

Comparison of GSR composition occurring at different locations around the firing position.

Abstract:  Variations in gunshot residue (GSR) compositions are used in the reconstruction of shooting incidents. In this study, GSR samples taken from seven different locations around and in the firearm were collected and analyzed using scanning electron microscopy/energy dispersive X‐ray analysis. Four different types of ammunition were applied. Very low correlations were found when different ammunition were used. This clearly shows that it is possible to differentiate between ammunition types. When the same ammunition was used, high correlations were found between samples taken from external positions (such as hands of shooter, bullet‐entrance holes) but poor correlation was found between internal samples (such as firearm barrel, cartridge case) and external samples. A high degree of association was found between samples that simulated victim and shooter. These findings clearly demonstrate that GSR comparison studies are meaningful but care needs to be taken when choosing suitable exhibits. External samples (such as hands of shooter, bullet‐entrance holes) are more suitable candidates than internal samples (barrel of the firearm, cartridge case).

Obtaining confidence intervals and Likelihood Ratios for body height estimations in images

In forensic practice, height estimations on perpetrators visible in video footage from surveillance cameras are regularly requested. There are several ways to do this. Insight is gained into the difference between actual and measured heights by taking validation measurements of a number of test persons. Variation between actual and measured heights is decomposed into a systematic part (because of height loss by pose, 3D modeling of the scene of crime, operator biases) and a random part (due to natural variation). On this basis a method is described for obtaining confidence intervals for the height, including head- and footwear, of questioned persons in images. Since the number of test persons is usually limited, the result is in terms of the Students t distribution. In addition, for cases in which a suspect is available, an expression is obtained for the Likelihood Ratio (LR), measuring the strength of evidence of resemblance of actual height of the suspect and measured height of the perpetrator. The Likelihood Ratio depends both on the rarity of the estimated perpetrators height and its closeness to the suspects height. Technical theorems included may be relevant for other forensic areas as well.

Posterior likelihood ratios for evaluation of forensic trace evidence given a two-level model on the data

In forensic science, in order to determine whether sets of traces are from the same source or not, it is widely advocated to evaluate evidential value of similarity of the traces by likelihood ratios (LRs). If traces are expressed by measurements following a two-level model with random effects and known variances, closed LR formulas are available given normality, or kernel density distributions, on the effects. For the known variances estimators are used though, which leads to uncertainty on the resulting LRs which is hard to quantify. The above is analyzed in an approach in which both effects and variances are random, following standard prior distributions on univariate data, leading to posterior LRs. For non-informative and conjugate priors, closed LR formulas are obtained that are interesting in structure and generalize a known result given fixed variance. A semi-conjugate prior on the model seems usable in many applications. It is described how to obtain credible intervals using Monte Carlo Markov Chain and regular simulation, and an example is described for comparison of XTC tablets based on MDMA content. In this way, uncertainty on LR estimation is expressed more clearly which makes the evidential value more transparent in a judicial context.

Variation in likelihood ratios for forensic evidence evaluation of XTC tablets comparison

In forensic comparison casework, where it is investigated whether items are from the same source, the likelihood ratio (LR) is a measure for evaluating the strength of evidence of the observed (dis)similarity between these items. The paper concentrates on evaluation of various LR models for comparison of XTC tablets from different production batches. Starting from a two‐level random effect model, where means are considered random, the distribution of these means may be assumed normal or approximated by kernel density estimations (KDEs), and variation in the data may or may not be allowed to differ per batch. It is investigated what effect the corresponding model and estimation choices have on the distribution of LRs for same‐ and different‐batch comparisons. Copyright

Possible influences on bullet trajectory deflection in ballistic gelatine

The influence of the distance to the top and bottom of a gelatine block and to bullet tracks from previously fired shots on a bullets trajectory, when passing through ballistic gelatine, was studied. No significant difference in deflection was found when trajectories of 9mm Luger bullets, fired at a 3.5cm distance to the top and bottom of a gelatine block and to bullet tracks from previously fired shots, were compared to trajectories of bullets fired 7cm or more away from any of the aforementioned aspects. A surprisingly consistent 6.5° absolute deflection angle was found when these bullets passed through 22.5 to 23.5cm of ballistic gelatine. The projection angle, determined by the direction of the deflection, appeared to be random. The consistent absolute angle, in combination with the random projection angle, resulted in a cone-like deflection pattern.

A sparse nearest mean classifier for high dimensional multi-class problems

The analysis of small datasets in high dimensional spaces is inherently difficult. For two-class classification problems there are a few methods that are able to face the so-called curse of dimensionality. However, for multi-class sparsely sampled datasets there are hardly any specific methods. In this paper, we propose four multi-class classifier alternatives that effectively deal with this type of data. Moreover, these methods implicitly select a feature subset optimized for class separation. Accordingly, they are especially interesting for domains where an explanation of the problem in terms of the original features is desired. In the experiments, we applied the proposed methods to an MDMA powders dataset, where the problem was to recognize the production process. It turns out that the proposed multi-class classifiers perform well, while the few utilized features correspond to known MDMA synthesis ingredients. In addition, to show the general applicability of the methods, we applied them to several other sparse datasets, ranging from bioinformatics to chemometrics datasets having as few as tens of samples in tens to even thousands of dimensions and three to four classes. The proposed methods had the best average performance, while very few dimensions were effectively utilized.

Practical Implementation of Blood Stain Age Estimation Using Spectroscopy

Blood is one of the most often encountered and valuable traces found at a crime scene. It not only contains information about what happened, but the chemical composition can also lead to the identification of the donor and even to the time of deposition, an important objective in forensic science. Many techniques have been explored for determination of the time of deposition, but none have made it into forensic practice. We have earlier used optical reflection spectroscopy for the age estimation of blood stains on white cotton, and to make this method suitable for forensic practice, we have adapted it to correct for the influence of the optical properties of the various substrates where blood may be found by extending a one-dimensional light-transport model to a two-layered model. With this model, we are able to calculate the relative amounts of OxyHemoglobin, MetHemoglobin, and Hemichrome in blood stains on colored surfaces, based on their reflectance spectra. We describe here a statistical method to calculate the 95% confidence interval around the calculated age. We demonstrate the applicability to field work and the potential value for crime investigations in a case study of a shooting incident. The absolute and relative ages of different groups of blood stains were measured using our new method. The measured age of blood stains can lead to a more realistic reconstruction of the timeline of events. To our knowledge, this paper presents the worlds first implementation in practice of a method for determining the age of blood stains.

Influence of Running on Pistol Shot Hit Patterns

In shooting scene reconstructions, risk assessment of the situation can be important for the legal system. Shooting accuracy and precision, and thus risk assessment, might be correlated with the shooters physical movement and experience. The hit patterns of inexperienced and experienced shooters, while shooting stationary (10 shots) and in running motion (10 shots) with a semi‐automatic pistol, were compared visually (with confidence ellipses) and statistically. The results show a significant difference in precision (circumference of the hit patterns) between stationary shots and shots fired in motion for both inexperienced and experienced shooters. The decrease in precision for all shooters was significantly larger in the y‐direction than in the x‐direction. The precision of the experienced shooters is overall better than that of the inexperienced shooters. No significant change in accuracy (shift in the hit pattern center) between stationary shots and shots fired in motion can be seen for all shooters.

Quantifying uncertainty in estimations of the total weight of drugs in groups of complex matrices.

In case practice at forensic drug departments, multiple items from one seizure are sometimes sent in with the question: what is the total amount of drugs in the seizure? This may be complicated especially if impregnated material is involved such as clothes or rubber. Measurement uncertainty is typically stable on drug percentages, not weights, and subsampling may take place. It is recognized more and more that determination and reporting of uncertainty on estimators are an essential part of obtaining scientifically sound results in the forensic field. Methodology is described to quantify uncertainty on estimations of the total drug weight in groups of complex matrices, given simple statistical models, along a subdivision of five types of cases. Given each of these types, case examples are given where uncertainty is quantified in estimations of drug weights, by means of confidence intervals. The described models are statistically sound and relatively easy to implement.

Correction to Tzidony and Ravreby (1992): A Statistical Approach to Drug Sampling: A Case Study

Abstract:  In 1992, Tzidony and Ravreby presented a confidence interval for the total weight of a seizure of illicit drugs present in a population. Their approach has subsequently been applied by several researchers in the field. The formula on which their approach is based does, however, not fully take into account the proportion of drug units found in the sample. In this paper, a modification is presented that consistently uses the correct sample size in all terms of the confidence interval, based on the proportion of drug units found in the sample. The effective sample size is smaller than the original sample size, and this should consequently be accounted for in the estimation of the standard error and in the corresponding t‐distribution. The new confidence interval is again based on the assumption that the proportion of drug units in the population is known after sampling.