数理统计Estimate.pptVIP

* * If the sample comes from a population which has a normal distribution then has a t-distribution there are different t-distributions for different sample sizes for estimating a single mean we use a t-distribution with n-1 degrees of freedom (df) they look like standard normal distributions but stretched out in the tails tn-1 approaches a normal distribution as n gets bigger. For n ≥ 30, there isn’t much difference. * Example: You want to estimate the average income of CEOs in large organizations. You take a sample of 15 companies and obtain a sample mean of $1.4m with a sample standard deviation of $0.3m. What is (a) your estimate of the population mean? (b) your 90% confidence interval for the mean? (c) your 95% confidence interval for the mean? * Example: Descriptive statistics: n = 15, = $1.4m, s = $0.3m. (a) your estimate of the population mean? $1.4m (b) your 90% confidence interval for the mean? 1.4 ± 1.761 × 0.3/?15 = 1.4 ± 0.14 ie, 90% CI is $1.26m to $1.54m (c) your 95% confidence interval for the mean? 1.4 ± 2.145 × 0.3/?15 = 1.4 ± 0.17 ie, 95% CI is $1.23m to $1.57m * Example: There have been complaints that the weights of bags of flour are typically lower than the stated 10kg. The producer denies this and claims the mean weight is as stated. A random sample of 100 bags gave a sample mean of 9.8kg with a SD of 1kg. Does this provide evidence that the complaints are justified? H0 : ? = 10 kg vs H1 : ? 10 kg Where m = mean weight of the bags of flour. Since we have a large random sample, the sampling distribution of the standardised sample mean will be standard normal. * t = Assuming a = 5%, we will reject H0 if t -1.645 (one-tailed to the left, since H1 says “ ”) Since the test statistic falls in the rejection region, we have sufficient evidence to reject H0 at a 5% significance level.It appears that, on average, the producer is slightly under