Hypothesis Testing Notes
Chapter Eight: Confidence Interval
Confidence Interval Basics
Definition: A method for estimating characteristics of a population based on sample data
Example Use: Estimating the average height of a population (e.g., US residents)
Confidence Level: Indicated as 99%, 98%, etc. (e.g., 99% confidence level means we are 99% certain that the true population parameter falls within the calculated interval)
Transition to Hypothesis Testing
Need for Hypothesis Testing
In situations where complete population data is unavailable, we utilize samples to draw conclusions about the population.
Process: Collect sample data, infer conclusions about population based on sample statistics.
Example: Testing a specific claim about population averages using sample data.
Hypothesis Definition
Hypothesis: A statement made regarding the characteristic of a population that we aim to test.
Example Hypothesis: "The average height of US residents is 5.11 feet."
Testing Hypotheses: Formulating claims (e.g., medication effect on high blood pressure) based on sampled data.
Insight into Medication Testing
Example of Testing: A researcher aims to find out if a new medication impacts high blood pressure.
Steps:
Sample a group of individuals and administer them the medication.
Analyze data to determine if there was a significant impact on blood pressure readings.
Importance of Probability: Statistical methods allow us to quantify the likelihood of results without resorting to mere guessing.
Understanding Hypothesis Testing
Purpose of Hypothesis Testing: To assess the likelihood of claims about population parameters.
Importance of Sample Evidence: Conclusions drawn from samples need to be tested for validity.
[Other tests in more advanced courses are available for further study]
Hypothesis Testing Procedure
Steps for Conducting Hypothesis Testing:
Make a statement regarding the nature of the population (null hypothesis).
Collect evidence (sample data).
Analyze evidence to assess plausibility, often expressed as a probability value.
Types of Hypotheses
Two Types of Hypotheses:
Null Hypothesis (H₀): Represents a statement of no effect/change. Assumed to be true until evidence suggests otherwise.
Example: "The drug has no effect on blood pressure."
Alternative Hypothesis (H₁ or Hₐ): Represents a statement that suggests there is an effect/change.
Example: "The drug has a positive effect on blood pressure."
Types of Hypothesis Tests
Types of Tests in Hypothesis Testing:
Two-tailed test: Testing for any significant difference (e.g., H₁: Not equal to a specific value).
Left-tailed test: Testing if the parameter is less than a specific value.
Right-tailed test: Testing if the parameter is greater than a specific value.
Writing Hypothesis Statements
Example Problem:
Research Question: Average working hours for employed males on weekends.
Null Hypothesis (H₀): H₀: μ = 5.46 hours (Assumed true)
Alternative Hypothesis (H₁): H₁: μ > 5.46 hours (Claim being tested)
Example Problem for Proportions:
Reported Preference: 55% of adults prefer name brand coffee.
Hypothesis:
Null Hypothesis (H₀): p = 0.55
Alternative Hypothesis (H₁): p ≠ 0.55
The test conducts a two-tailed assessment based on whether the preference differs from 55%.
Understanding Conclusions in Hypothesis Testing
Conclusion of Hypothesis Testing:
Two Possible Outcomes:
Reject the null hypothesis (H₀).
Do not reject the null hypothesis (H₀).
Language of Conclusions:
Conclusively rejecting H₀ means the evidence supports H₁.
Insufficient evidence leads to non-rejection of H₀.
Accuracy of statistical claims is vital. Conclusions state probabilities rather than absolute truths.
School Example: Mean Length of Phone Calls
Null Hypothesis: H₀: μ = 3.525 minutes.
Alternative Hypothesis: H₁: μ < 3.525 minutes.
Upon calculating: Rejecting H₀ indicates support for H₁, stating there is sufficient evidence to conclude mean phone call duration is less.
Summary of Language Used in Conclusions:
Favoring H₁: "There is sufficient evidence to support H₁."
Favoring H₀: "There is not enough evidence to support H₁."
Ethical Considerations
Ethical Communication in Statistics:
Data claims should be presented with caution regarding their accuracy.
Example Commentary: Commercial evidence presented should clarify claims based on evidence collected, rather than asserting unqualified statements.