Abstract
The ability to identify human remains is an important focus in the forensic sciences. The use of modern techniques, such as dental records and DNA comparisons, may be rendered inaccurate as a result of decomposition, or finding fragmentary or completely skeletised remains. As a result of this, other techniques needed to be found. One such technique is the use of discriminant function analysis; a technique which assists in identifying the sex of unknown remains.
This dissertation investigated the use of discriminant function analysis to sex the upper limb long bones of n=219 black South Africans from the Raymond A. Dart Collection housed at the University of the Witwatersrand, Johannesburg. The analysis made use of 33 linear measurements for each individual, 10 on the humerus, 11 on the radius and 12 on the ulna. The application of choice most often used by researchers to analyse statistical data is IBM’s Statistical Package for the Social Sciences (SPSS). However, this statistical package is expensive to access. Therefore, this study used Microsoft Excel (Excel) to run the analysis. The findings were validated by using SPSS and rerunning the analysis. This was done to determine if future researchers could rely solely on Excel to avoid the costs associated with SPSS.
The findings showed that the analysis of the data by means of Excel enables one to distinguish males and females. The accuracy of the results from this study ranged from 65.0% to 86.8% and are in line with the findings from previous research. Of the three long bones, the radius was found to be the most discriminatory by 0.5%. A combination of nine linear measurements, excluding those which are highly correlated to another, gave an accuracy result of 86.8%. The highest rate for the humerus of 86.3% was achieved through a combination of the five linear measurements which were found to be significant at p<0.05. The ulna distinguished males and females correctly 86.2% of the time when all linear measurements, except length, were combined. When considering a univariate analysis, the single best distinguishing measurement was found to be the
v
mediolateral diameter of the ulna midshaft at 84.3%. The transverse diameter of the proximal epiphysis provided the highest accuracy rate of 83.1% for the humerus, while the mediolateral diameter of the distal epiphysis gave the highest accuracy rate of 80.8% for the radius. From a univariate perspective, the ulna was found to be the most discriminatory, yet from a multivariate perspective, it was the radius. Overall results from this study showed a range of 65.% to 86.3% for the humerus; 67.1% to 86.8% for the radius; and 65.0% to 86.2% for the ulna.
It can be concluded that the long bones of the upper limb can accurately sex individuals in a black South African population, using a discriminant analysis run in Excel. It is worthwhile to note that a recent study has identified a positive secular trend regarding stature amongst urban black South Africans. Given that the Dart Collection is comprised of mainly rural migrant workers, one should be mindful of the potential impacts that secular trends may have on future applications.
Keywords: Sex Determination, Upper Limb, South Africa, Discriminant Function Analysis, Microsoft Excel, IBM SPSS.