ABSTRACT
Many techniques in forensic anthropology employ osteometric data, although little work has been done to investigate the intrinsic error in these measurements. These data were collected to quantify the reliability of osteometric data used in forensic anthropology research and case analyses. Osteometric data (n = 99 measurements) were collected on a random sample of William M. Bass Donated Collection skeletons (n = 50 skeletons). Four observers measured the left elements of 50 skeletons. After the complete dataset of 99 measurements was collected on each of the 50 skeletons, each observer repeated the process for a total of four rounds. The raw data is available on Mendeley Data ( DCP Osteometric Data, Version 1. DOI: 10.17632/6xwhzs2w38.1). An example of the data analyses performed to evaluate and quantify observer error is provided for the variable GOL (maximum cranial length); these analyses were performed on each of the 99 measurements. Two-way mixed ANOVAs and repeated measures ANOVAs with pairwise comparisons were run to examine intraobserver and interobserver error, and relative and absolute technical error of measurement (TEM) was calculated to quantify the observer variation. This data analysis supported the dissemination of a free laboratory manual of revised osteometric definitions (Data Collection Procedures 2.0[1], pdf available at https://fac.utk.edu/wp-content/uploads/2016/03/DCP20_webversion.pdf) and an accompanying instructional video (https://www.youtube.com/watch?v=BtkLFl3vim4). This manual is versioned and updatable as new information becomes available. Similar validations of scientific data used in forensic methods would support the ongoing effort to establish valid and reliable methods and protocols for proficiency testing, training, and certification.
ABSTRACT
This study evaluates the reliability of osteometric data commonly used in forensic case analyses, with specific reference to the measurements in Data Collection Procedures 2.0 (DCP 2.0). Four observers took a set of 99 measurements four times on a sample of 50 skeletons (each measurement was taken 200 times by each observer). Two-way mixed ANOVAs and repeated measures ANOVAs with pairwise comparisons were used to examine interobserver (between-subjects) and intraobserver (within-subjects) variability. Relative technical error of measurement (TEM) was calculated for measurements with significant ANOVA results to examine the error among a single observer repeating a measurement multiple times (e.g. repeatability or intraobserver error), as well as the variability between multiple observers (interobserver error). Two general trends emerged from these analyses: (1) maximum lengths and breadths have the lowest error across the board (TEM<0.5), and (2) maximum and minimum diameters at midshaft are more reliable than their positionally-dependent counterparts (i.e. sagittal, vertical, transverse, dorso-volar). Therefore, maxima and minima are specified for all midshaft measurements in DCP 2.0. Twenty-two measurements were flagged for excessive variability (either interobserver, intraobserver, or both); 15 of these measurements were part of the standard set of measurements in Data Collection Procedures for Forensic Skeletal Material, 3rd edition. Each measurement was examined carefully to determine the likely source of the error (e.g. data input, instrumentation, observer's method, or measurement definition). For several measurements (e.g. anterior sacral breadth, distal epiphyseal breadth of the tibia) only one observer differed significantly from the remaining observers, indicating a likely problem with the measurement definition as interpreted by that observer; these definitions were clarified in DCP 2.0 to eliminate this confusion. Other measurements were taken from landmarks that are difficult to locate consistently (e.g. pubis length, ischium length); these measurements were omitted from DCP 2.0. This manual is available for free download online (https://fac.utk.edu/wp-content/uploads/2016/03/DCP20_webversion.pdf), along with an accompanying instructional video (https://www.youtube.com/watch?v=BtkLFl3vim4).
