91Ó°ÊÓ

Statistical hypothesis testing is a method used to make decisions about the properties of a population based on sample data. When we conduct a hypothesis test, we typically start with a null hypothesis, which is a statement of no effect or no difference. In our case, the null hypothesis (H0) could be 'Yawning is not affected by exposure to yawning,' meaning the proportion of people who yawn when exposed to yawning is the same as when not exposed.

The alternative hypothesis (H1) would be 'Yawning is affected by exposure to yawning,' meaning the proportion of people who yawn when exposed to yawning is different from when they are not exposed. The goal is to gather enough evidence from the sample data to either reject H0 in favor of H1 or fail to reject H0, keeping H0 as a plausible conclusion based on the available data.

Common steps in hypothesis testing include:

• Formulating H0 and H1
• Collecting sample data
• Calculating a test statistic based on the sample
• Determining a P-value to evaluate the test statistic
• Making a conclusion based on the P-value and a pre-determined significance level (like 0.05)

By following these steps, researchers can make objective decisions supported by data rather than intuition or speculation.

The P-value is a probability that measures the evidence against the null hypothesis. A smaller P-value indicates stronger evidence against H0, suggesting that the observed data is unlikely under H0.

To calculate the P-value, we first need a test statistic that summarizes the sample data. For example, in the MythBusters' yawning experiment, we used a z-test because we are comparing proportions between two groups. The test statistic (z) is calculated based on the sample proportions and pooled proportion.

For our example:

• Proportion for exposure group (p1) = 10/34
• Proportion for non-exposure group (p2) = 4/16
• Pooled proportion (p) = (10 + 4) / (34 + 16) = 0.28
• Test statistic (z) = \( \frac{p1 - p2}{\text{sqrt}(p \times (1 - p) \times (1/t1 + 1/t2))} \) = 0.69

After calculating the z-value, we refer to standard normal distribution tables to find the P-value. The resulting P-value lets us determine if we have enough evidence to reject the null hypothesis. In this case, a high P-value suggests we do not have strong evidence to claim that yawning is contagious.

Fisher's Exact Test is a statistical test used to determine if there are nonrandom associations between two categorical variables. It's particularly useful when sample sizes are small, as it doesn't rely on the assumptions of larger samples.

In the context of the MythBusters' experiment, we are comparing the 'yawners' and 'non-yawners' in both the exposed and non-exposed groups. Fisher's Exact Test calculates the exact probability of obtaining the observed results, or more extreme, under the null hypothesis.

Steps include:

• Arranging data in a 2x2 contingency table:
  
  • Exposed and yawned: 10
  • Exposed and did not yawn: 24
  • Not exposed and yawned: 4
  • Not exposed and did not yawn: 12
• Calculating the probability of each possible outcome
• Summing these probabilities to find the P-value

In our example, Fisher's Exact Test gave a P-value of 0.5128, which is quite high. This indicates there is no significant association between yawning and exposure to yawning based on the sample data. The conclusion supports the null hypothesis, suggesting yawning might not be as contagious as traditionally believed.

Proportion testing involves comparing proportions to understand if differences observed between groups are statistically significant. In the MythBusters' experiment, we compare the yawning proportions of two groups: those exposed to yawning and those not exposed.

Here’s a quick guide to the steps involved in proportion testing:

• Determine sample proportions (p1 and p2)
• Calculate the pooled proportion (p)
• Compute the standard error of the difference between proportions
• Find the test statistic (z-value)
• Look up the P-value for the test statistic

In the MythBusters' example:

• p1 = 10/34 for the exposed group
• p2 = 4/16 for the non-exposed group
• p = 0.28 (pooled)
• Standard error = \(\text{sqrt}(p \times (1 - p) \times (1/t1 + 1/t2))\)
• z = 0.69
• P-value = corresponds to z = 0.69

A high P-value indicates no significant difference in proportions. Thus, based on the data, yawning exposure does not significantly increase the probability of yawning in others.