Introduction to Chapter 10 Concepts

TL;DR

Chapter 10 introduces hypothesis testing, a crucial statistical method for making decisions about a population based on sample data. You'll learn how to set up competing claims (hypotheses) and use sample evidence to decide which claim is more likely true. The core idea is to see if your sample results are unusual enough to reject a default assumption about the population.

1. The Mental Model

Imagine you're a detective. Someone claims a coin is fair, but you suspect it's biased. You flip it 100 times and get 70 heads. Is that enough evidence to say it's biased, or could a fair coin just randomly do that? Hypothesis testing is the formal way to answer questions like this.

2. The Core Material

Chapter 10 is all about the process of hypothesis testing. This process allows us to use data from a sample to draw conclusions about an entire population. It's fundamentally about making a decision: either we have enough evidence to support a new idea, or we stick with the status quo.

2.1. Hypotheses: The Competing Claims

Photo by Asad Photo Maldives on Pexels

Every hypothesis test starts with two opposing statements about a population parameter (like a mean, proportion, or standard deviation):

• Null Hypothesis ($H_0$): This is the "status quo" or the default assumption. It usually states that there is no effect, no difference, or no change. Think of it as "innocent until proven guilty." We assume $H_0$ is true until we have strong evidence against it.
  • Example: The average height of adult men is 69 inches ($H_0: \mu = 69$).
• Alternative Hypothesis ($H_a$ or $H_1$): This is what you're trying to prove. It contradicts the null hypothesis. It states there is an effect, a difference, or a change.
  • Example: The average height of adult men is not 69 inches ($H_a: \mu \neq 69$). Or, it could be greater than 69 ($H_a: \mu > 69$), or less than 69 ($H_a: \mu < 69$).

2.2. The Logic: Evidence Against the Null

Photo by Google DeepMind on Pexels

We gather sample data and calculate a test statistic. This statistic measures how far our sample result is from what we'd expect if the null hypothesis were true.

Then, we calculate a p-value. The p-value is the probability of observing a sample result as extreme as, or more extreme than, what we actually got, assuming the null hypothesis is true.

• Small p-value (e.g., less than 0.05): This means our observed sample result would be very unlikely if the null hypothesis were true. This provides strong evidence against the null hypothesis, so we reject $H_0$.
• Large p-value (e.g., greater than 0.05): This means our observed sample result isn't particularly unusual if the null hypothesis were true. We fail to reject $H_0$. This doesn't mean $H_0$ is true, just that we don't have enough evidence to say it's false.

2.3. Significance Level ($\alpha$)

Photo by Fez Brook on Pexels

Before we even collect data, we decide on a significance level ($\alpha$). This is the threshold for our p-value. Common values are 0.05 (5%) or 0.01 (1%). If p-value $\leq \alpha$, we reject $H_0$. If p-value $> \alpha$, we fail to reject $H_0$.

Here's the general flow:

graph TD
    A["Formulate Null (H0) and Alternative (Ha) Hypotheses"] --> B["Choose Significance Level (alpha)"];
    B --> C["Collect Sample Data"];
    C --> D["Calculate Test Statistic"];
    D --> E["Determine P-value"];
    E --> F{{"Is P-value <= alpha?"}};
    F -- Yes --> G["Reject H0"];
    F -- No --> H["Fail to Reject H0"];
    G --> I["Conclude Ha is supported"];
    H --> J["Conclude Ha is not supported (insufficient evidence)"];

3. Worked Example

Let's say a company claims its new light bulbs last 1000 hours on average. You suspect they last less than 1000 hours. You decide to test this claim.

1. Hypotheses:
   • $H_0: \mu = 1000$ hours (The average lifespan is 1000 hours)
   • $H_a: \mu < 1000$ hours (The average lifespan is less than 1000 hours)
2. Significance Level: You choose $\alpha = 0.05$.
3. Collect Data: You randomly sample 50 bulbs and find their average lifespan is 980 hours, with a sample standard deviation of 70 hours.
4. Calculate Test Statistic & P-value: (This step involves specific formulas you'll learn, but for this intro, let's assume the calculation leads to...)
   • ...a test statistic (e.g., a t-score) of -2.0.
   • ...which gives a p-value of 0.025.
5. Decision: Compare p-value to $\alpha$: $0.025 \leq 0.05$.
   • Since the p-value (0.025) is less than $\alpha$ (0.05), you reject $H_0$.
6. Conclusion: You have sufficient evidence to conclude that the average lifespan of the new light bulbs is less than 1000 hours.

4. Key Takeaways

• Hypothesis testing lets you make informed decisions about a population based on sample data.
• The null hypothesis ($H_0$) is the status quo, and the alternative hypothesis ($H_a$) is what you're trying to prove.
• The p-value tells you how likely your sample results are if $H_0$ were true.
• If the p-value is small (less than $\alpha$), you reject $H_0$ because your results are too unlikely under $H_0$.
• If the p-value is large (greater than $\alpha$), you fail to reject $H_0$ because your results aren't unusual enough.
• Rejecting $H_0$ means you have evidence for $H_a$; failing to reject $H_0$ means you don't have enough evidence for $H_a$.

Common Mistakes to Avoid:
* Assuming $H_0$ is true when you fail to reject it: "Failing to reject" isn't the same as "accepting" $H_0$. It just means you don't have enough proof to say it's false.
* Confusing the p-value with the probability $H_0$ is true: The p-value is about the data given $H_0$, not the probability of $H_0$ itself.
* Not setting up hypotheses correctly: Make sure $H_0$ always includes an equality (=, $\leq$, or $\geq$) and $H_a$ includes an inequality ($\neq$, $<$, or $>$).
* Changing $\alpha$ after seeing the p-value: Decide your significance level before you analyze your data.

5. Now Try It

Think of a claim you've heard about a group of people or objects (e.g., "The average commute time in my city is 30 minutes," or "More than 50% of people prefer coffee over tea").
1. Formulate a null hypothesis ($H_0$) and an alternative hypothesis ($H_a$) for this claim.
2. State what your conclusion would be if you got a very small p-value (e.g., 0.001) and what it would be if you got a very large p-value (e.g., 0.60), assuming you set $\alpha = 0.05$.

Success looks like clearly stating $H_0$ and $H_a$ using appropriate symbols, and correctly explaining the decision to reject or fail to reject $H_0$ based on the p-value relative to $\alpha$.