E-balonmano.com: Revista de Ciencias del Deporte (Jun 2016)
Analysis of data quality and generalizability observing system counter in elite handball
Abstract
An ad hoc tool observation system comprising a field formats and a set of mutually exclusive and exhaustive categories (E/ME) is presented for studying counterattack in handball. Data quality has a qualitative aspect faced through the agreed matching. The quantitative aspect of this analysis is performed using the Kappa index Cohen, three correlation coefficients Kendall Tau b, Pearson and Spearman and analysis of Generalizability. For these quantitative analyzes sequential analysis program SDIS-GSEQ and generalizability program SAGT v were used. 1.0. Generalizability analysis not only allows to determine the reliability of observers, but also estimate the goodness of fit of the categories and optimize the design of measurement to calculate the minimum number of sessions needed to accurately generalize the results of research. The results confirm that observational tool allows for reliable, accurate and valid records. It has also estimated the homogeneity of taxonomic system as an E/ME system. Design optimization of measurement calculated that six parties needed to achieve accurate generalization of results; determines high reliability of observers.