A CT-scan database for the facial soft tissue thickness of Taiwan adults

Facial soft tissue thicknesses: Noise, signal, and P

“…For example: occlusal line in normal body mass index individuals differs by 1.7 mm (p = 0.014) between the sexes or 9 % of the mean occlusal line value [6]; mid-masseter differs by 3.0 mm (p = 0.008) between the sexes or 11 % of the mean mid-masseter value [6]; lower lip margin differs by 4.16 mm (p = 0.014) between Taiwanese and Turkish adults aged 18-29 years or 35 % of the mean lower lip margin value [19]; and gonion differs by 4.40 mm (p = 0.000) between South African and American Black adults or 28 % of the mean gonion depth [8].…”

“…When used in this manner, almost any difference that is not nil will register as being statistically significant, eliminating any value of P over and above the mean. P is also not an indicator of the strength of a difference, so reporting it to many decimal places (see e.g., [5,6,8,19,26]) is of little benefit. The P-value does not encode any practical meaning of the difference [86,87], it simply communicates that the difference is not nil [85].…”

“…Some recent examples will help illustrate: statistically significant sex difference of 0.93 mm (p=0.009) reported at infra-orbital [6]; statistically significant sex difference of 1.0 mm (p=0.000) reported at nasion [26]; statistically significant difference between Taiwanese and Turkish adults of 0.79 mm (p=0.000) reported at supraglabella [19]; statistically significant difference between Turkish and White Europeans of 0.94 mm (p=0.00) reported at lateral nasal [5]; and statistically significant difference between South African Blacks and American Blacks of 0.33 mm (p=0.006) reported at midphiltrum [8].…”

“…A number of factors promote this popularity: i) facial soft tissue thicknesses (FSTT) can be relatively quickly and simply measured; ii) central tendency description via the calculation of arithmetic means is enticingly straightforward; and iii) extensive options exist for categorizing samples to facilitate publication. These qualities have driven a recent explosion of papers in the field [3]; just in Forensic Science International alone there have, for example, been no less than 16 contributions between 2009 and mid-2015 [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19].…”

“…Despite prior calls to caution that small statistically significant differences should not be automatically interpreted as biological signal [1-3, 20, 21], recent studies show a sustained hesitation to abandon past practice both in regards to sex and ancestry distinction (see e.g., [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][22][23][24][25]). It seems it is difficult to refrain from assigning practical significance to statistically significant test results, even when the mean differences are diminishingly small.…”