APPLICATION OF DIFFERENT STRATEGIES FOR DATA QUALITY CONTROL OF AN OBSERVATION TOOL FOR YOUNG SOCCER

Abstract

Ibon Etxeazarra Escudero, Julen Castellano Paulis y Oidui Usabiaga Arruabarrena

This work is contextualized in the field of young soccer and aims
to estimate the data quality of an observation tool which is ad hoc designed using
two strategies: Kappa coefficients and Generalizability Theory. This instrument
(SOFBAS) is formed by nine requirements and 73 categories, contemplating
several aspects related to players, team, and interaction. After 14 observers follow
a training process using an observation protocol, a record was conducted to
estimate the data quality. Taking the obtained records as the starting point, an
estimation of the coefficients of the Cohen and Fleiss’ Kappa was carried out,
and the study of the components of variance was applied to the study of
generalizability. The obtained results, association values above 0.8 and accuracy
in the generalization close to 1 for observer’s facet, allow us to analyze data
quality from different strategies, ensuring greater accuracy of the assessment.
To conclude, the suggested tool is reliable to proceed in a systematic way to
coding and game action including in young soccer.