STAT-S 300 Lecture Notes - Lecture 26: Bias Of An Estimator, Confidence Interval, Sampling Distribution
Document Summary
12-18-13: choose srs of size n1 from population 1 with mean and standard deviation, choose independent second srs of size n2 from population 2 with mean and standard deviation. Shape- when population distribution normal, sampling distribution of is normal. If population shape unknown, is normal when n1 > 30 and n2 > 30 by. Graph if n1 or n2 < 30, normal if no strong skew/no outliers. Difference in sample means is unbiased estimator of difference in population means. Spread- standard deviation = as long as n < 1/10n. Two-sample t statistics: for rare cases when and known, use two-sample z statistic as basis for inference about, when and unknown, use two-sample t statistic- has approximate (not exact) t distribution. Estimate and by standard deviations from samples (s1 and s2) Option 1 (technology)- use t distribution with df calculated from somewhat messy formula: may not come out as whole number, (do not use)
