Letters to the Editor / British Journal of Oral and Maxillofacial Surgery 52 (2014) 870–873 873

Accuracy and reliability of linear measurements using 3dimensional computed tomographic imaging software for

Le Fort I osteotomy

We thank the authors of the previous letter about our paper for sharing their valuable knowledge and experience in statistical

The intrac be used when that are organ units in the sa used to quant tionship (suc of a quantita cation is the quantitative m are measurin

The intrac ance within th interviewers. as it approac iners. As it a examiners. T total variance between inte

An impor calculated as sum of error between subj from a hetero seem to be hi from a group characteristic it is applied to a group with a narrow range of the same characteristic.

We understand and agree with the author’s letter that there are many statistical methods available to test validity and reliability, but we strongly disagree that the method chosen was insufficie nd tha nce t PE, F y. Psych ennan 993;47 r R, B cients.

E-mai (B.F. (O.S. felip .doi.orgmethods. lass correlation is a descriptive statistic that can quantitative measurements are made on units ised into groups, and it describes how strongly me group resemble each other.1 It is commonly ify the degree to which subjects with a fixed relah as full siblings) resemble each other in terms tive trait (such as heritability). Another appliassessment of consistency or reproducibility of easurements made by different observers who g the same quantity. lass correlation is the proportion of the total varie data that is explained by the variance between

In most cases the value ranges from 0-1, where hes 1 we see perfect agreement between exampproaches zero we see no agreement between he intraclass correlation is the proportion of the within the data that is explained by the variance rviewers.1–3 tant issue is bias. Any reliability coefficient is the ratio of variance between subjects to the variance and subject variance. If the variance ects is sufficiently high (that is, the data come geneous sample) then reliability will inevitably gh. If the intraclass correlation is applied to data of subjects with a wide range of the measured , reliability will seem to be higher than when letter a

Refere 1. Shrou abilit 2. MacL

Stat 1 3. Mülle coeffi http://dxnt for our research. However, we appreciate the nk the authors for their constructive remarks. leiss JL. Intraclass correlations: uses in assessing rater reliol Bull 1979;86:420–8.

RN. Interrater reliability with SPSS for Windows 5.0. Am :292–6. üttner P. A critical discussion of intraclass correlation

Stat Med 1994;13:2465–76.

Bruno Felipe Gaia ∗

Lucas Rodrigues Pinheiro

Otávio Shoite Umetsubo

Oséas Santos Júnior

Felipe Ferreira Costa

Marcelo de Gusmão Paraíso Cavalcanti

Dental School, University of São Paulo, Brazil ∗ Corresponding author. l addresses: bgaia@usp.br, gaia90@gmail.com

Gaia), lpr@usp.br (L.R. Pinheiro), osu@usp.br

Umetsubo), oseasjunior@usp.br (O.S. Júnior), ecosta@usp.br (F.F. Costa), mgpcaval@usp.br (M. de Gusmão Paraíso Cavalcanti)

Available online 14 June 2014 /10.1016/j.bjoms.2014.05.008