Comment on “Evaluation of Manual and Two-Rotary NiTi Retreatment Systems in Removing Gutta-Percha Obturated with Two Root Canal Sealers”

Proper research design, appropriate evaluation and measurement methods, use of correct statistical tests, interpretation, and inference are the essentials without which any scientific research reporting is incomplete and does not serve its proposed purpose. In this communication, some common flaws in research designing, evaluation, analysis, and inference have been explained using a published article as a reference. The purpose of the paper is to help the scientific community recognize the significance of proper research designing and planning in order to achieve the results which are worthwhile. With reference to the research article entitled “Evaluation of manual and two-rotary NiTi retreatment systems in removing Gutta-Percha obturated with two root canal sealers” [1], we would like to point out certain queries related to research design, statistical and inferential errors. Scoring Method. The authors have shown scoring criteria of debris in Section under the heading Analysis. The criteria seem to be observer dependent. Such methods can be highly subjective in nature and hence are not reliable unless they are proven unanimously. As scores are ordinal in nature and the criteria for calculation being highly subjective, it is difficult to assume that these criteria are replicable unless otherwise proven. The authors have quoted a reference for using the criteria [2]; however, we are sorry to state that the source article too does not justify the use of the criteria as a reliable tool. For observations needing subjective evaluation, reliability and replicability are the most essential requirements [3]. The contemporary literature supports achieving multiple-observer agreement under such situations [4, 5]. Such revalidation is essential as ordinal scoring pattern varies considerably owing to minor change in subjective evaluation. For example, observer 1 might evaluate 49% debris in a case while the same case might be evaluated as 51% by observer 2, thus scoring them as 2 and 3, respectively, which on an ordinal scale with a 5-point scale shows 20% variation in results. Statistical Tools. It is difficult to digest that the authors have used scoring pattern using an ordinal scale (nonparametric in nature) but have analyzed the results using parametric tests (ANOVA and Newman-Keuls test). Even if they have used parametric test (ANOVA), choice of Newman-Keuls test seems to be inappropriate as it tends to increase the Type I error. Although the whole point of multiple comparison post hoc tests is to keep the chance of a Type I error in any comparison to be 5%,