Sample size and power calculation for univariate case in quantile regression

Yanuar, Ferra Sample size and power calculation for univariate case in quantile regression. IOP Conf. Series: Journal of Physics: Conf. Series.

Text ICE STEM 2018. Sample size and power Yanuar_2018.pdf Download (984kB)

Abstract

The purpose of this study is to calculate the statistical power and sample size in simple linear regression model based on quantile approach. The statistical theoretical framework isthen implemented to generate data using R. For any given covariate and regression coefficient, we generate a random variable and error. There are two conditions for error distributions here; normal and nonnormal distribution. This study resulted that for normal error term, sample size is large if the effect size is small. Meanwhile, the level of statistical power is also affected by effect size, the more effect size the more level of power. For nonnormal error terms, it isn’t recommended using small effect size, moderate effect size unless sample size more than 320 and large effect size unless sample size more than 160 because it resulted in lower statistical power.

Actions (login required)

View Item