2018 tallied facial soft tissue thicknesses for adults and sub-adults

Abstract

The tallied facial soft tissue thicknesses (or T-Tables) represent grand means of published facial soft tissue thickness sample means. These sample means have been drawn from across the full-breadth of the facial soft tissue thickness (FSTT) literature, including forensic science, anthropology and odontology. The report of new summary statistics for >1290 new sub-adults and >2200 new adults since the last T-Table calculation, in 2008 for sub-adults and 2013 for adults respectively, makes their update timely. The maximum sample sizes at any landmark now stand at 3023 for individuals aged 0–11 years old (g–g′); 3145 for individuals aged 12–17 years old (n–se′); and 10,333 for adults (n–se′). Following the recalculation of grand weighted means and comparison to the original 2008 data, some shifts in the T-Table statistics are evident at specific landmarks, namely: 2–2.5mm increases at gonion (go–go′) and mid-mandibular border (mmb–mmb′) for adults; 3.5mm decrease at gonion (go–go′) for 12–17year olds; and 2.0mm decrease at menton (me–me′) for 0–11year olds. Differences at all other landmarks (91–100% depending on the dataset) were minimal being <1.0mm. Performance tests of the new grand means as point estimators (using individuals with known FSTT size from the C-Table), show the 2018 T-Table statistics to produce marginally less error than the 2013 means: 2018 standard error of the estimate=3.7mm in contrast to 2013 standard error of the estimate=3.9mm. The long run nature of the T-Table statistics (i.e., big data) and quantified performance test accuracies on known subjects, earmark the 2018 T-Table as the premier FSTT standard for craniofacial identification casework. In the distant future, this is likely to change as the C-Table raw data repository grows, allowing shorths and shormaxes to be calculated for large samples. Given current raw data repository sample sizes of 0–1574 for T-Table landmarks (notably lower for younger individuals), there is some way to go before enhanced central tendency estimators can entirely replace untrimmed arithmetic means.