THE INTERVAL NATURE OF AN ORDINAL SCALE FOR MEASURING THE MARGINAL FRACTURE OF AMALGAM

The clinical performance of amalgam alloys over time has been assessed by measuring the extent of marginal fracture of restorations made fro m these alloys. Scales of photographs of restorations exhibiting varyi ng degrees of marginal fracture have been used to make these assessmen ts. The purpose of this study was to determine the relationship betwee n the ordinal scale numbers of five photographs composing a commonly u sed scale (Mahler and Marantz, 1979) and the average width in micromet ers of marginal fracture of the restorations in the photographs of thi s scale. In addition, a comparison was made between parametric and non -parametric statistical methods when applied to marginal fracture data . The results showed that four of the five photographs of this scale d emonstrated a significant linear regression with marginal fracture wid th (R2 = 0.997; p = 0.002). The last photograph of the scale, which pr oved to be an outlier, was not used in the regression and was accommod ated by an extrapolation procedure. Using previously gathered clinical data on the marginal fracture behavior of five amalgam alloys, the us e of parametric statistical procedures (ANOVA and Scheffe's multiple c omparison test) proved to be more discriminatory than the use of non-p arametric procedures (Kruskal-Wallis and Dunn's Multiple comparison te st) when tested at the same overall confidence level. Thus, having a p hotographic scale of an interval nature removes any doubt about using the more powerful technique of parametric statistics to evaluate the m arginal fracture behavior of dental amalgams. In addition, based on th ese data in the form of marginal fracture widths and Maryniuk's (1990) survey of dentists' replacement decisions, a restoration having a mar ginal fracture width of approximately 100 mum or more, is at risk of b eing replaced.