Discriminant analysis in treatment evaluation of ectopic eruption of the maxillary first permanent molars.

Abstract

The value of an analytical routine which could be used in the evaluation of clinical studies was assessed. Stepwise discriminant analysis with stepwise selection/elimination was used to reduce the amount of data without losing information before further analyses with canonical discriminant analysis. The purpose of canonical discriminant analysis was to reveal interactions between factors associated with the ectopic eruption of the first permanent molars. Two canonical discriminant functions were used to determine the accuracy of diagnosis and the effect of treatment. This information consisted of variables which were the most important (their ranks) and their single correlations (structure coefficients) to the function itself. This information was interpreted clinically. The coefficients of the canonical discriminant functions can be used for classification purposes. The use of a scatterplot instead of simple classification tables is suggested. The scatterplot was particularly informative on how well the procedure discriminated between groups. A table based on original data is included to allow the reader to test the validity of the findings on cases of his/her own. This multivariate strategy avoids spurious intercorrelations between two or more variables analysed as multiple single factors that has a high risk of erroneous conclusions.