Stat Hypothesis

Introduction to Confidence Intervals and Hypothesis Tests

  • There are four different types of confidence intervals and eventually seven different types of hypothesis tests.

  • Similarities between confidence intervals and hypothesis tests outweigh differences.

  • Hypothesis testing is conceptually more challenging than confidence intervals.

  • Understanding these concepts is crucial for future coursework.

Confidence Intervals

  • Used to estimate an unknown parameter.

  • Examples include calculating the average height of a population based on a sample.

Hypothesis Testing

  • Also known as significance testing.

  • Utilized to make decisions based on incomplete data.

  • Important for understanding probability and decision-making in life.

Conceptual Framework

  • Hypothesis testing is essentially about making binary decisions based on evidence.

  • Recognizes the importance of error and the consequences of mistakes in decision-making.

Real-World Example: The Jury System

  • A case example of a person accused of a crime (e.g., Akhil) going to trial.

  • The jury's decision is binary: convict or acquit.

  • The system relies on a preset default known as the null hypothesis.

  • Both defense and prosecution present their evidence, but the burden of proof lies on the prosecution.

Legal Standards

  • Presumption of Innocence: The default outcome wherein the accused is considered innocent until proven guilty.

    • Jury convicting only when evidence surpasses a certain threshold (proof beyond a reasonable doubt).

  • Definition of Reasonable Doubt: A high percentage of confidence—approx. 90% and subject to debate.

Comparison of Criminal and Civil Cases

  • Criminal Case: Brought by the district attorney, requires proof beyond a reasonable doubt.

  • Civil Case: Involves lawsuits, and the evidence only needs to show a preponderance.

    • Example: O.J. Simpson's criminal trial vs. subsequent civil trial.

Understanding Burden of Proof

  • The concept of burden of proof can be associated with an 'alpha level' in statistics.

  • The null hypothesis (H0) indicates the presumed outcome—default belief that no effect/relationship exists.

    • The alternative hypothesis (Ha) opposes this, suggesting an effect or relationship.

Hypothesis Structure in Trials

  • In criminal cases:

    • Null hypothesis: not guilty.

    • Alternative hypothesis: guilty.

  • Requires convincing evidence to reject the null hypothesis in favor of the alternative.

Hypothetical Scenario with Cards

  • Classroom activity involving drawing a card to illustrate concepts.

  • Example setup: Students draw cards; if black, they win; if red, the teacher wins.

  • Introduces the idea of considering whether the outcomes indicate fair play or cheating.

Arguments from Legal Perspectives

  • Defense argues luck based on distribution of outcomes (e.g., if results were close to expected).

  • Prosecution argues that drastically uneven outcomes (e.g., winning 11 of 15) indicate foul play.

  • Null hypothesis in this scenario: the deck is fair (50/50 chance).

Probability and Decision Making

  • Probability of attaining observed results under the null hypothesis considered (p-value).

  • A lower p-value suggests that observed results are unlikely under the assumption of the null hypothesis being true.

    • Example in the card situation: A 6% chance of being that lucky indicates insufficient evidence to convict.

  • Conclusion for juries: Default to the null hypothesis unless evidence proves otherwise.