91Ó°ÊÓ

Hypothesis testing is a method used in statistics to decide whether a certain belief or claim about a population parameter is supported by evidence collected from a sample. In our exercise, we want to determine if two types of steel have the same average fracture toughness using a two-sample z test. We start by setting up two hypotheses: the null hypothesis (H_0 ) and the alternative hypothesis (H_a ).

• The null hypothesis (H_0 ) is a statement of no effect, suggesting that both types of steel have the same average fracture toughness, namely \( \mu_1 = \mu_2 \).
• The alternative hypothesis (H_a ) contrasts H_0 and proposes that the averages are different, i.e., \( \mu_1 e \mu_2 \).

Hypothesis testing uses sample data to assess the evidence against the null hypothesis. If the data is highly unlikely under H_0, we may conclude that H_a is more likely true. The formal decision is made by comparing the calculated p-value to a pre-decided significance level (usually 0.05). If the p-value is smaller, we reject H_0 , indicating the evidence supports the alternative hypothesis.

The p-value is a crucial part of hypothesis testing, representing the probability of observing results as extreme as, or more extreme than, those measured, assuming the null hypothesis is true. In our exercise, we calculate the p-value for each sample size (n=100 and n=400) using the following steps:

• Calculate the test statistic ( z -score), which indicates how many standard deviations away the sample mean difference is from zero under H_0 .
• For instance, for n = 100 , the z -score is 1.414. For n = 400 , it's 2.828.
• Using the standard normal distribution, find the area in the tails beyond the calculated z -score. This is done through statistical tables or software.
• The p-value for n = 100 is about 0.157, and for n = 400 , it's approximately 0.0047.

P-value helps assess whether the observed data is consistent with H_0. Lower p-values suggest that the data is unlikely under H_0, hence providing evidence in favor of H_a. However, it doesn't express the size or importance of the difference found.

Statistical significance is about determining whether an observed effect could reasonably occur by chance. It's commonly assessed by comparing the p-value against a significance level, often set at 0.05.

• If the p-value is less than the significance level, the result is deemed statistically significant, suggesting strong enough evidence to reject the null hypothesis.
• In our steel example, with n = 100 producing a p-value of 0.157, we do not find statistical significance.
• Yet for n = 400 , a p-value of 0.0047 indicates statistical significance, allowing us to reject the null hypothesis.

However, statistical significance does not imply that the findings are practically important or large. It simply says that the effect observed is unlikely to have arisen merely by chance. Thus, it's crucial to further investigate whether the difference is meaningful in real-world terms.

Practical significance evaluates whether the size of an effect is large enough to be of value in a practical sense, rather than merely being statistically significant.

• In our example, while the difference in fracture toughness was statistically significant for n = 400 with a p-value of 0.0047, the actual difference of 0.2 units in average fracture toughness is too minor to matter practically.
• It's essential to consider whether the observed difference is relevant in a real-world context, such as engineering standards or cost efficiency.

Practical significance helps us understand the real-world impact of our findings beyond the mathematical computations of statistical significance. So, simply reporting the p-value without context might paint an incomplete picture. Practical implications should always accompany statistical results for a comprehensive evaluation.