Evaluation of the correlation box (Segner) for characterization of the facial skeleton

The convenience of the so-called harmonious combinations of five basic radiographic cephalometric measurements (Segner), derived from multiple regressions and presented in the form of a correlation box, as a means of diagnosing skeletal malocclusion has been evaluated statistically. The study involved 55 untreated orthodontic patients 7-14 years old with Angle's Class I malocclusion. The correlation box was built stepwise on the basis of four linear regressions computed between five angular measurements (SNA, SNB, NSL/ML, NSL/NL, and NSAr). The regression analyses indicated that the correlation box was not accurate enough. The coefficients of determination for the regressions varied between 0.26 (for NSL/NL) and 0.63 (for SNB). Most part of the variation in NSL/ML, NSL/NL, and NSAr angles remained unexplained. Individual norms indicated by the box for the ANB angle seemed doubtful in many cases. The correlation box seemed to show the statistically most probable associations between measurements used. The combinations may be harmonious but the existence of other harmonious combinations cannot be excluded.