91Ó°ÊÓ

A confidence interval gives us a range where we expect the true population parameter, like a mean, to fall. It doesn't just rely on a single number but gives a possible range of values that estimate an unknown parameter. For instance, the 95% confidence interval implies that if you were to take 100 different samples and compute a confidence interval for each one, approximately 95 of them would contain the true mean.

When calculating a confidence interval for a small sample, it's crucial to use the t-distribution rather than the normal distribution. This is because the t-distribution accounts for extra variability in the sample mean when the sample size is small. In our exercise, the confidence interval was computed using a t-value and the formula:

• \[ \bar{x} \pm t_{n-1, \alpha/2} \left( \frac{s}{\sqrt{n}} \right) \]
• where \( \bar{x} \) is the sample mean.
• \( s \) is the standard deviation.
• \( n \) is the sample size, and \( \alpha/2 \) is the alpha level for the interval.

A confidence interval offers insights into the stability and reliability of the sample mean as an estimator for the population mean.

While a confidence interval provides an estimate for the mean of the population, a prediction interval gives a range within which we can expect an individual future observation to fall. Prediction intervals are wider than confidence intervals because they account for both the variability of the sample mean and the variability of individual outcomes.

The formula for a prediction interval is similar to that for a confidence interval but includes an additional term to capture the added uncertainty:

• \[ \bar{x} \pm t_{n-1, \alpha/2} \cdot s \cdot \sqrt{1 + \frac{1}{n}} \]
• Here, \( \sqrt{1 + \frac{1}{n}} \) adjusts for the extra individual observation variability.

This adjustment makes the prediction interval broader, acknowledging that individual measurements vary more than the average does. In practical settings, such as predicting an individual's test score, using a prediction interval helps address these uncertainties.

A Q-Q (quantile-quantile) plot is a graphical tool to help assess whether a dataset comes from a particular distribution, like the normal distribution. It compares the quantiles of the sample data with the quantiles of a normal distribution. If the points on the plot lie approximately along a straight line, the data is likely normally distributed.

To create a Q-Q plot:

• Plot points corresponding to the sorted data values on one axis.
• Plot the expected quantiles of the normal distribution on the other axis.

Any major deviations from a straight line indicate departures from normality.

Using a Q-Q plot is especially useful because it provides a visual assessment of data normality, which can be more intuitive than numerical tests. It does not only tell you whether data adheres perfectly to normal distribution, but it also indicates where departures occur, making it a helpful diagnostic tool.

The Shapiro-Wilk test is a statistical test that determines how well sample data fit a normal distribution. It is one of the most powerful tests for checking normality. Unlike graphical methods like the Q-Q plot, which provide a visual assessment, the Shapiro-Wilk test gives a formal statistical measure that can support decision-making about normality.

The test operates by calculating a W-statistic:

• If the W-statistic is close to one, the data is very likely to be normally distributed.
• A low W-statistic suggests deviations from normality.
• The test provides a p-value, and a small p-value (typically less than 0.05) indicates that the null hypothesis — that the data is from a normal distribution — is inconsistent with the observed data.

For samples like the one in the exercise, using the Shapiro-Wilk test helps to quantify the assumption of normal distribution, providing a solid statistical basis for further inference.