Category Archives: Forensic Statistics

Forensic anthropologists — lend me your data

Please note: This is not a new post, but a restored post from August that I lost in a WordPress upgrade

Friends, Romans, forensic anthropologists, lend me your data

I have been reading the Journal of Forensic Sciences (JFS) over the last couple of days to see what sort of research is being done in forensic science and to see how many studies are using statistics to make, or to reinforce their conclusions. The answer the the second question is “quite a few.” There has been quite significant adoption of multivariate analysis, most commonly PCA, in a wide variety of forensic disciplines fields and that is very pleasing to see.

Forensic anthropology

Anthropologists, and in particular forensic anthropologists have long been heavy users of statistical methodology. Many studies use linear regression, linear discriminant analysis, principal component analysis and logistic regression. The well-known and widely used forensic anthropology computer programme FORDISC uses LDA. It is interesting to me to see the appearance of some newer/different classification techniques such as k-nearest neighbour, quadratic discriminant analysis, classification and regression trees, support vector machines, random forests, and neural networks.

Forensic anthropology features heavily in JFS, and the papers contain a large amount of statistical analysis of data. The focus of the articles is often on classification of remains in to age, gender, or racial groups, or on age estimation. The articles are generally quite interesting and well written.
Show me the data

However, there is almost never any provision of the raw data, and my experience whilst writing my data analysis book was that there was no response to my requests for data from even a single anthropologist of the dozen or so that I wrote to. Not even a polite “sorry but we are unable to release the data.” I understand that in all scientific disciplines data can be expensive, in terms of time or money, to collect, and so a researcher might justifiably want to retain a data set as long as possible to get as much research value from it as possible. However, surely there must be a point where the data could be released in the public domain? The University of Tennessee Knoxville does have a forensic anthropology databank, but, at least from the webpage, it seems that there is an emphasis on deposits rather than withdrawals.

A challenge

I therefore issue a challenge to the forensic anthropology community – release some of your data into the wild. It will benefit your discipline as it will others, and you might find your work cited more as people give you credit for producing the data that they are applying their novel techniques to.

Share Button