The language of hypothesis testing

Welcome to the World of Hypothesis Testing!

Ever heard a claim and thought, "I bet that’s not actually true"? Maybe a friend claims they can predict a coin toss 80% of the time, or a company says their new seeds grow faster. In Statistics, we don't just guess; we use Hypothesis Testing to decide if there is enough evidence to back up these claims.

Don't worry if this seems like a lot of new vocabulary at first. Think of it like a courtroom trial: we assume someone is innocent (the "status quo") unless we have enough evidence to prove otherwise. Let’s break down the language you need for your OCR A Level exams.

1. The Two Hypotheses: \( H_0 \) and \( H_1 \)

Every test starts with two competing statements. We always write these in terms of a parameter (for this chapter, that’s usually \( p \), the probability of success in a binomial distribution).

The Null Hypothesis (\( H_0 \)): This is the "boring" version. It assumes nothing has changed and the situation is exactly as it was before. We always use an equals sign here.
Example: \( H_0: p = 0.5 \) (The coin is fair).

The Alternative Hypothesis (\( H_1 \)): This is the "exciting" version. It’s the claim we are testing for. We use symbols like \( > \), \( < \), or \( \neq \).
Example: \( H_1: p > 0.5 \) (The coin is biased towards heads).

Quick Tip: Defining the Parameter

OCR examiners love it when you define your symbols! Always write: "where \( p \) is the population proportion of..." right after your hypotheses.

Key Takeaway: \( H_0 \) is the "innocent" assumption (equality), and \( H_1 \) is the "guilty" suspicion (inequality).

2. The Test Statistic and Significance Level

The Test Statistic: This is simply the result you get from your sample. If you flip a coin 20 times and get 15 heads, "15" is your test statistic. It is the evidence we use to judge \( H_0 \).

The Significance Level (\( \alpha \)): This is the "bar" we set for evidence. Common levels are 5% (\( 0.05 \)) or 1% (\( 0.01 \)).
If the probability of our result happening by pure chance is less than this level, we say the result is "significant" and reject \( H_0 \).

Did you know?

The significance level is actually the probability of incorrectly rejecting the null hypothesis. It’s the risk we take of being wrong!

Key Takeaway: The test statistic is the evidence, and the significance level is the threshold we use to judge that evidence.

3. One-Tailed vs. Two-Tailed Tests

How do we decide which direction to look in? It depends on what the question asks.

1-Tail Test: We are looking for a change in a specific direction.
Keywords: "increased", "decreased", "better", "worse".
Example: \( H_1: p > 0.2 \) or \( H_1: p < 0.2 \).

2-Tail Test: We are looking for any change at all, regardless of direction.
Keywords: "has changed", "is different".
Example: \( H_1: p \neq 0.2 \).

Common Mistake to Avoid:

In a 2-tail test at the 5% level, you must split the significance level in half: 2.5% for the "too high" end and 2.5% for the "too low" end!

Key Takeaway: Use 1-tail for specific directions and 2-tail when you just want to see if a value has changed.

4. Critical Values and the Critical Region

Think of the Critical Region (or Rejection Region) as the "Red Zone." If our test statistic falls inside this zone, it is so unlikely to have happened by chance that we reject \( H_0 \).

Critical Value: The "border" number that starts the critical region.
Acceptance Region: The range of values where we don't have enough evidence to change our minds, so we stick with \( H_0 \).

Analogy: The High Jump

The Critical Value is the height of the bar. The Test Statistic is how high you actually jumped. If you jump higher than the bar (fall into the critical region), you’ve done something impressive enough to prove the "status quo" wrong!

Key Takeaway: If your result is in the Critical Region, you reject \( H_0 \). If not, you "fail to reject" \( H_0 \).

5. The P-value

The p-value is the probability of getting a result as extreme as, or more extreme than, the one we actually observed, assuming \( H_0 \) is true.

Simple Rule:
- If p-value \( \leq \) Significance Level \( \rightarrow \) Reject \( H_0 \).
- If p-value \( > \) Significance Level \( \rightarrow \) Do not reject \( H_0 \).

Memory Trick:

"If the P is low, the Null must go. If the P is high, the Null can fly (stays)."

Key Takeaway: The p-value tells us how likely our result was. A tiny p-value means our result was very rare, suggesting \( H_0 \) might be wrong.

6. Writing the Perfect Conclusion

OCR is very strict about how you word your final answer. You must never say you have "proven" anything. Statistics is about evidence, not 100% certainty.

The "Two-Part" Conclusion:

Part 1: Statistical Result
State whether you reject \( H_0 \) or not, and mention the significance level.
"There is evidence at the 5% level to reject \( H_0 \)..."

Part 2: Contextual Conclusion
Link it back to the story in the question.
"...suggesting that the proportion of people who support the new law has increased."

Wrong Wording (Do NOT use these!):

- "I accept \( H_0 \)" (We only "fail to reject" it).
- "This proves the coin is biased" (It only "suggests" or provides "evidence").

Key Takeaway: Always be tentative. Use words like "evidence," "suggests," and "likely."

Quick Review Box

- Null Hypothesis (\( H_0 \)): The "no change" statement (always uses \( = \)).
- Alternative Hypothesis (\( H_1 \)): The "change" statement (uses \( >, <, \text{ or } \neq \)).
- Significance Level: The threshold for evidence (e.g., 5%).
- Critical Region: The range of values that leads you to reject \( H_0 \).
- P-value: The probability of your result happening by chance.