Estimacion bootstrap para el coeficiente de determinacion: un estudio de simulacion

The application of bootstrap approach allows the estimations of the measurements of precision and carry out contrast of hypothesis in situations in which we haven't got information about the sample distribution of one statistic or in situations in which the sample distribution is dependent on unknown parameters. This is the case for the determination coefficient, the most usual index in the empirical context used to evaluate the adjustment for the linear regression model. The aim of this study is to explore how well this procedure works in the estimation of the measurement error of R2, and also, in the determination of confidence intervals using the percentile method and the Bca method This is achieved by applying Monte Carlo simulation