Reproducibility of Hypothesis Testing and Confidence Interval

摘要: Abstract P -value is the probability of observing a current sample and possibly other samples departing equally ormore extremely from null hypothesis toward postulated alternative hypothesis. When p less thana certain level called (= 0 : 05), researchers claim that supported empirically.Unfortunately, some findings discovered in way are not reproducible, partly because itself isa statistic vulnerable to random variation. Boos Stefanski (2011) suggests calculating upper limitof testing, using bootstrap predictive distribution. To determine sizeof replication study, this study proposes thought experiments by simulating boosted samplesof different sizes given observations. The method illustrated for cases two-group comparisonand multiple linear regression. This also addresses reproducibility points 95%confidence interval. Numerical examples show center point covered 95% confidence intervalsgenerated resamples. However, end with 50% chance. Hence studydraws graph rate each parameter interval.Keywords: Reproducibility,