A comparison of approximate confidence interval methods for standardized effect sizes in two-independent and two-dependent samples : tukey's lamda distribution
Abstract:
The objective of this research were find out an approximate confidence interval methods that make have a least confidence interval. The approximate confidence interval methods for standardized effect size in two-dependent and two-independent samples: tukeys lamda distribution for 5 difference skewness (0.25, 0.5, 1.0, 1.5 and 2.0), kurtosis (2, 4 ,6 ,8 ,10, 12 and 14), confidence coefficient (0.90, 0.95 and 0.99) and sample size (2, 4, 6, 8, 12, 16, 24, 28, 32, 48, 56, 64, 96, 112 and 128). For each statistics under experimental situation, the Monte Carlo experiment was repeated 3,000 times. The findings were summarized as follows. 1. Case two-independent samples the gH and dH methods have a least confidence interval in all of situations and be equal length. 2. Case two-dependent samples the dU method have a least confidence interval when skewness = 2, kurtosis=10. 3. When increase a length confidence interval will decrease. 4 .When sample sizes are large the dB, gU, dU, gL1, dL1, gL2, dL2, gH and dH method in a confidence interval have a length were decrease except gB method have a length were increased when sample sizes is large.