Detailed Study Notes on Air Quality Index Analysis and Hypothesis Testing

Analysis of Air Quality Index and Hypothesis Testing

The Air Quality Index (AQI) is subject to fluctuations year over year. It can be higher or lower compared to previous years, giving rise to uncertainty in mean values across time.
Historical mean AQI values are compared to assess current observation significance.

Null Hypothesis (H0): There is no significant difference in the mean AQI from a specified value; notably, H0 states that the mean AQI is equal to 120 (the historical mean over the past five years).
Alternative Hypothesis (H1): There is a significant difference, implying the mean AQI is not equal to 120. This hypothesis implies a testable change in air quality.
Students should be capable of expressing these hypotheses both statistically and in verbal terms.

Data is collected by taking 10 random AQI measurements in February. This number (n = 10) is crucial for subsequent statistical analyses.
For example, suppose the observed mean AQI from these measurements is 105, with a standard deviation of 6.2.

To determine if the sample mean (105) is significantly different from the null hypothesis mean (120), a statistical test is employed.
Question: How likely is this observed mean if the null hypothesis is true? This involves calculating a test statistic and determining a p-value.

The test statistic is calculated as:
- t = \frac{\bar{x} - \mu_0}{s / \sqrt{n}}
- Where:
  - \bar{x} = observed sample mean (105)
  - \mu_0 = mean under null hypothesis (120)
  - s = standard deviation of sample (6.2)
  - n = sample size (10)
Plugging in the values:
- t = \frac{105 - 120}{6.2 / \sqrt{10}}
- This yields a t-value of -7.9.

Determine the likelihood of encountering a t-value of -7.9 on Student's t-distribution.
Calculate the p-value reflecting this observation given the null hypothesis is true. A low p-value indicates the null hypothesis should be rejected, while a higher p-value suggests it should not.
General rules: If p < alpha (commonly 0.05), reject H0; if p > alpha, do not reject H0.

The t-distribution resembles the standard normal distribution but has thicker tails.
As sample size (n) grows larger, the t-distribution approaches the shape of a normal distribution, benefitting from properties of both distributions.
T-values are evaluated relative to critical t-values based on degrees of freedom (df) determined by df = n - 1 .
- In the example, df = 10 - 1 = 9 .

When computing p-values, both tails of the distribution must be accounted for. Use Excel's T.DIST formula to automatically compute the two-tailed p-value.
Important: Use absolute t-values in calculations; the T.DIST function does not accept negative t-values.
For one-tailed tests, use the T.DIST.RT function for right tail analysis.

For a two-tailed test at an alpha level of 0.05 with nine degrees of freedom, the critical t-values would be approximately ±2.262.
Observing a t-value of -7.9 means it falls beyond the critical values, confirming rejection of the null hypothesis.

Type I Error: Rejecting the null hypothesis when it is actually true; the probability of making this error is controlled by the alpha level.
Type II Error: Failing to reject the null hypothesis when it is false. It’s important to note that these errors indicate uncertainty rather than affirming the truth of the null hypothesis.

One-sided tests are only conducted when the direction of the effect is specified (e.g., higher or lower) and the alternative hypothesis is directional.
Two-sided tests are generally preferred as they assess the possibility of any significant difference, regardless of direction.
It is critical to define whether to conduct a one-sided or two-sided test before analyzing data as it influences p-value interpretations significantly.

Confidence intervals provide a range around the sample estimate to infer whether a hypothesized parameter (like the mean of the population) falls within this interval.
The construction of confidence intervals hinges on the same sample statistics (mean and standard deviation) used in hypothesis tests.