THE ACCURACY DISTORTIONS IN SELF-ASSESSMENT

Abstract

This article presents the results of two algorithms analysis to verify how clear and sound could be a particular personality self-assessment. The first algorithm validated the convergence between different multidimensional personality constructs by applying the nomological network method. Central network criteria are betweenness, centrality degree, and extended influence at the matrix. On the other hand, a parallel hierarchical cluster analysis specifies the proximity between personality measures divided into two types of “strengths” and “weaknesses”. The second analytic algorithm applied is focused on socially desirability biases. By classifying twelve groups with results that are below, fit to or are above the normal distribution (M ± 1SD) compared to K-means centering coefficients two types of ranges were investigated using Receiver Operating Characteristics (ROC) parameters. This method was applied to assess the probability of false-positive/false-negative measurement errors. The ecological validation of the ROC analysis is also elaborated by considering “vignettes” of the eight high-schoolers that took part in the experimental self-assessment. Their final SDR scores and the measurement error risk was distributed helping evaluate the probability rate results to be overly self-enhancing or misleading