91Ó°ÊÓ

In statistical analysis, the **normality assumption** is a key concept that underlies many statistical methods, including confidence intervals and hypothesis testing. This assumption states that the data follows a normal distribution, often visualized as the classic bell curve. Ensuring that your data meets this assumption is critical, as many statistical tests are only valid when this condition is satisfied.
Visual inspection techniques like Q-Q plots are often used, where data points are plotted against a theoretical normal distribution. If the points lie approximately along a straight line, it's an indication that the data is normally distributed.
However, visual inspection can sometimes be subjective. For a more objective test of normality, the Shapiro-Wilk test is often employed. This statistical test provides a test statistic and a p-value, where a p-value greater than 0.05 generally suggests that the data does not significantly deviate from a normal distribution.

The **Shapiro-Wilk test** is a statistical test that helps in determining whether a dataset follows a normal distribution. It is particularly useful for smaller sample sizes, like the one in our exercise, which involves 15 metal rods.
Here's how it works:

• First, the test computes a test statistic which ranges between 0 and 1.
• Next, it provides a p-value, which is key in making a decision about normality.

A p-value greater than 0.05 often supports the assumption of normality, suggesting that any deviation from the normal distribution is statistically insignificant. This means that the data is potentially appropriate for further parametric testing, which depends on normality assumptions.
For example, if you conduct the Shapiro-Wilk test on the rod diameters in this exercise and find a p-value of 0.20, the assumption of normality stands. This result allows us to proceed with methods like the t-distribution for calculating confidence intervals.

The **t-distribution** is crucial in statistics, especially when dealing with small sample sizes, typically less than 30. It is similar to the normal distribution but is slightly wider or has fatter tails, which accounts for increased variability when estimating a population parameter from a small sample.
Unlike the standard normal distribution, the t-distribution is defined by degrees of freedom (df), calculated as the sample size (n) minus one. For instance, with 15 rods, you have 14 degrees of freedom.
When calculating a confidence interval, the t-distribution helps determine the range in which we expect the population mean to fall. For a two-sided 95% confidence interval, you would find the critical t-value that corresponds to 2.5% in each tail (the remaining 5% is divided equally due to two tails).

• If you’re calculating only an upper bound, like in the exercise, you focus solely on the upper tail with 5% significance.

Hence, the calculations differ slightly, leading to distinct results for two-sided intervals and single bounds.