91Ó°ÊÓ

In hypothesis testing, the null hypothesis is a key concept. It's essentially an initial assumption or statement that there is no effect or difference in the context of the problem being studied. For example, the null hypothesis

• Often symbolized as \( H_0 \) • Makes a baseline claim about a population parameter, like saying two groups are equal

In our example of comparing on-time flight rates between two airlines, the null hypothesis (\( H_0 \)) assumes both airlines have the same on-time rates, i.e., \( p_A = p_B \).
The main goal of hypothesis testing is to gather enough evidence to determine whether or not to reject this starting assumption. Depending on the data and test results, the null hypothesis can either be rejected in favor of an alternative hypothesis, or not rejected, suggesting there's insufficient evidence to make a new conclusion.

In hypothesis testing, a critical value helps in deciding whether to reject the null hypothesis. It's a threshold that the test statistic is compared against and serves as a marker dividing regions of acceptance from rejection.
Here's what you need to know:

• The critical value is determined by the chosen significance level \( \alpha \), which indicates the probability of rejecting the null hypothesis when it's true • In a two-tailed test like ours, critical values occur at both extremes of the distribution

Given an \( \alpha = 0.01 \) significance level, we look up the critical values using a z-table for a two-tailed test. For our example, the critical values are \( z = \pm 2.576 \). This means if the calculated test statistic falls below \( -2.576 \) or above \( 2.576 \), we reject the null hypothesis and opt for the alternative one.

The standard error (SE) is an essential concept in statistics. It measures the variability or dispersion of a sampling distribution. In simpler terms, it tells us how much the sample statistic is expected to vary from the actual population parameter. The smaller the standard error, the closer the sample mean is to the true population mean.
Here's how to understand it:

• Standard error is based on sample size and data variability • It's calculated differently depending on the data type and test

For our example, we needed the standard error of the difference in proportions between two airlines. The pooled proportion (\( \hat{p} \)) was calculated first and then used alongside sample sizes of Airline A and B to find SE:
\[SE = \sqrt{\hat{p}(1 - \hat{p})\left(\frac{1}{n_A} + \frac{1}{n_B}\right)} \approx 0.0433.\]This SE value helped us determine the variability from which we calculated the test statistic, enabling a comparison against the critical value.

The test statistic is a standardized value that measures the degree of deviation of the sample statistic from the null hypothesis. It’s a crucial part of hypothesis testing as it quantifies the evidence against the null hypothesis. The process involves comparing this statistic to a critical value to decide whether to reject the null hypothesis.
Here's a breakdown:

• A higher or lower test statistic relative to the critical value can lead to rejecting the null • Test statistics vary by type of test (z, t, chi-square)

In our situation, we calculated a z-test statistic for the difference in flight on-time proportions:\[z = \frac{(\hat{p}_A - \hat{p}_B)}{SE} = \frac{(0.71 - 0.74)}{0.0433} \approx -0.692.\]This z-value represents how many standard errors the observed difference of sample proportions is away from the hypothesized difference of zero. In our case, \( -0.692 \) did not reach the critical region for \( z \), so we did not reject the null hypothesis.