Letter to the Editor Response to “Significance of descriptive statistics in forensic anthropology research”, published in the November 2013 issue of the Journal of Forensic and Legal Medicine Dear Sir, The authors of a recent letter 1 have pointed to the omission of descriptive statistics in a paper authored by us in early 2013. 2 However, we are surprised that they were “astonished” by the omission, 1 particularly since several of the “basic details” of descriptive statistics which they allude to (standard error, inter- quartile range, results of a normality test) are missing in a research paper on stature estimation co-authored by one of them. 3 Now, agreed that descriptive statistics are able to provide some useful information such as the measures of central tendency and mea- sures of dispersion of the study sample. Nevertheless, descriptive statistics perhaps take on greater importance in studies whose objective is to categorise or differentiate two groups (e.g., sex dif- ferentiation) where the mean and standard deviations of various parameters in each of the groups can prove highly valuable to researchers. But in the case of predicting a dependent variable (e.g., stat- ure, age) based on corresponding independent variable(s) (e.g., tooth dimensions; tooth mineralisation) using statistical analyses such as least squares regression or ridge regression, the statisti- cal indicators mentioned by the authors may assume lower levels of importance. The degree of correlation (expressed by correla- tion coefficients) and its level of significance, as well as error rate in predicting the dependent variable (e.g., the Standard Error of the Estimate for the sample; the Mean Absolute Difference) appear to be the more commonly reported statistical mea- sures. 3e5 This is also our experience of publishing research related to regression analysis in leading forensic journals, including the Journal of Forensic and Legal Medicine, 6e9 wherein the peer review process barely noticed the omission of descrip- tive statistics d possibly an indication of its diminished practical importance in such studies. What is ironic though is that, in their attempt to point out the statistical ‘deficiencies’ in our paper, 2 the authors of the letter 1 have taken a few statistical missteps of their own. They seem to believe that descriptive statistics is the panacea for gauging poten- tial bilateral variation as well as sex differences in tooth dimen- sions. This is evident in their statement, “It would have been interesting to note if the size of teeth varied significantly between sides. Basic descriptive statistics would have given a reasonable idea to these questions.” The authors seem to have forgotten a basic lesson in statistics d that descriptive statistics aims to summarise a sample, and the appropriate statistical analysis to determine whether ‘significant’ differences exist between the right and left sides of the dentition are inferential statistics such as tests of differ- ences (e.g., Paired Samples t-test). Also, whether the correlation to stature of the teeth on the two sides varied was less of a concern since the ridge regression attempted to determine the correlation of the dentition, as a whole (i.e., as a unit), to stature. Next, the authors deviate from their comment concerning the ‘significance’ (sic) of descriptive data and begin to exhort the importance of sex-specific data and how descriptive statistics can, again, assist in providing “the answers to these questions” d ques- tions on whether there are differences between males and females for stature and tooth dimensions. Of course, differences are ex- pected to exist between the male and female dentitions (as well as malefemale stature). However, researchers would be keener to apply, yet again, tests of differences (e.g., Independent Samples t-test) since descriptive statistics do not give the statistical signifi- cance of the difference. Moreover, we note the continual use of the word “significance” by the authors of the letter (in the title and two other instances in the text); perhaps they are unaware that this constitutes non- technical usage of a technical term in statistics and should have been avoided. Lastly, the authors’ comment that combining male and female data “can be the reason for lower levels of significance and higher errors in estimate” contravenes the basic tenets of research, which is the keen observation of and logical deduction from facts. We are astonished that the authors did not think it worthy of contacting us first to verify unpublished results at our disposal before attempting an inference. As it turns out, assessing the sexes separately continued the trend of low correlation (r  0.40) which was mostly statistically insignificant. Yours sincerely, Drs. Sudeendra Prabhu, Ashith B. Acharya, and Mahadevayya V. Muddapur. Ethical approval None. Funding None. Conflict of interests None. DOI of original article: http://dx.doi.org/10.1016/j.jflm.2013.09.001. Contents lists available at ScienceDirect Journal of Forensic and Legal Medicine journal homepage: www.elsevier.com/locate/jflm Journal of Forensic and Legal Medicine 22 (2014) 148e149 1752-928X/$ e see front matter Ó 2014 Elsevier Ltd and Faculty of Forensic and Legal Medicine. All rights reserved. http://dx.doi.org/10.1016/j.jflm.2013.12.025