9.1 Basic concepts and definitions

Hypothesis testing is a crucial tool in statistical analysis, allowing researchers to make informed decisions based on data. This process involves formulating null and alternative hypotheses, which represent competing claims about a population parameter.

Understanding the types of errors that can occur in hypothesis testing is essential. Type I errors (false positives) and Type II errors (false negatives) have different implications and trade-offs, influencing how we interpret and apply statistical results in real-world scenarios.

Null vs Alternative Hypotheses

Defining Hypotheses in Statistical Testing

Top images from around the web for Defining Hypotheses in Statistical Testing

• 

  Hypothesis Testing (4 of 5) | Concepts in Statistics View original

  Is this image relevant?

• 

  Comparing two means – Learning Statistics with R View original

  Is this image relevant?

• 

  Introduction to Hypothesis Testing | Concepts in Statistics View original

  Is this image relevant?

• 

  Hypothesis Testing (4 of 5) | Concepts in Statistics View original

  Is this image relevant?

• 

  Comparing two means – Learning Statistics with R View original

  Is this image relevant?

1 of 3

Top images from around the web for Defining Hypotheses in Statistical Testing

• 

  Hypothesis Testing (4 of 5) | Concepts in Statistics View original

  Is this image relevant?

• 

  Comparing two means – Learning Statistics with R View original

  Is this image relevant?

• 

  Introduction to Hypothesis Testing | Concepts in Statistics View original

  Is this image relevant?

• 

  Hypothesis Testing (4 of 5) | Concepts in Statistics View original

  Is this image relevant?

• 

  Comparing two means – Learning Statistics with R View original

  Is this image relevant?

1 of 3

• Null hypothesis (H₀) represents no effect or difference, typically maintaining status quo
• Alternative hypothesis (H₁ or Hₐ) contradicts null hypothesis, proposing research claim or effect
• Hypothesis testing makes inferences about population parameters using sample data
• Null hypothesis assumed true until evidence suggests otherwise (principle of parsimony)
• Testing determines if sufficient evidence exists to reject null hypothesis for alternative
• Hypotheses must be mutually exclusive and exhaustive, covering all possible outcomes
• Formulate hypotheses before data collection to avoid bias in testing process

Role and Importance of Hypotheses

• Null hypothesis serves as baseline for comparison in statistical analysis
• Alternative hypothesis represents researcher's expectation or theory to be tested
• Well-formulated hypotheses guide experimental design and data collection methods
• Clear hypotheses help in selecting appropriate statistical tests (t-test, ANOVA)
• Hypotheses facilitate interpretation of results and drawing meaningful conclusions
• Proper hypothesis formulation ensures scientific rigor and reproducibility of research
• Examples of hypotheses in different fields:
  • Psychology: H₀: There is no difference in memory recall between young and old adults H₁: Young adults have better memory recall than old adults
  • Economics: H₀: Changes in interest rates do not affect inflation rates H₁: Changes in interest rates affect inflation rates

One-tailed vs Two-tailed Tests

Characteristics and Differences

• One-tailed tests examine relationship in one direction, two-tailed tests consider both directions
• One-tailed tests place entire α (significance level) in one tail of distribution
• Two-tailed tests split α equally between both tails of distribution
• One-tailed tests have greater power to detect effect in hypothesized direction
• Two-tailed tests more conservative, used when no strong prior expectation about effect direction
• Choice between one-tailed and two-tailed tests made a priori, justified by research question and literature
• Examples of test types:
  • One-tailed: Testing if a new drug increases patient survival rates
  • Two-tailed: Investigating if a change in diet affects blood pressure (increase or decrease)

Considerations for Test Selection

• One-tailed tests appropriate when previous research or theory suggests specific directional effect
• Two-tailed tests used when exploring potential differences without directional hypothesis
• Misuse of one-tailed tests (post-hoc selection) can inflate Type I error rates
• One-tailed tests cannot detect effects in opposite direction of hypothesis
• Two-tailed tests provide more comprehensive analysis of potential effects
• Consider ethical implications of test choice in fields like medicine or public policy
• Examples of appropriate use:
  • One-tailed: Testing if a new teaching method improves test scores based on previous studies
  • Two-tailed: Investigating if a new fertilizer affects crop yield without prior knowledge of direction

Type I vs Type II Errors

Understanding Error Types

• Type I error occurs when null hypothesis incorrectly rejected when actually true (false positive)
• Type II error occurs when null hypothesis not rejected when actually false (false negative)
• Probability of Type I error denoted by α (alpha), also significance level of test
• Probability of Type II error denoted by β (beta), 1 - β represents power of test
• Type I and Type II errors inversely related; decreasing one typically increases other
• Examples of errors:
  • Type I: Concluding a medical treatment is effective when it actually isn't
  • Type II: Failing to detect a harmful environmental pollutant that actually exists

Implications and Considerations

• Type I errors lead to wasted resources, incorrect scientific conclusions, potential harm in medicine
• Type II errors result in missed opportunities, failure to detect important effects, delayed progress
• Balancing Type I and Type II errors crucial in research design and interpretation
• Strategies to minimize errors include:
  • Increasing sample size to improve power and reduce both error types
  • Setting appropriate significance levels based on field standards and research goals
  • Using more precise measurement techniques to reduce variability in data
• Consider consequences of errors in specific contexts (clinical trials, environmental monitoring)

Significance Level, Critical Value, and p-value

Defining Key Concepts

• Significance level (α) predetermined threshold for rejecting null hypothesis (typically 0.05 or 0.01)
• Critical value point on distribution separating rejection region from non-rejection region
• P-value probability of obtaining results at least as extreme as observed, assuming null hypothesis true
• P-value less than significance level (p < α) leads to rejection of null hypothesis
• P-value greater than or equal to α fails to reject null hypothesis
• Critical value approach and p-value approach equivalent methods for statistical decisions
• Examples of significance levels in different fields:
  • Psychology research often uses α = 0.05
  • Medical studies may use more stringent α = 0.01 for critical health decisions

Relationships and Applications

• Visualize concepts on sampling distribution: critical value defines rejection region boundary
• P-value represents area beyond observed test statistic on distribution
• As sample size increases, critical value tends to decrease, allowing detection of smaller effects
• Relationship between concepts guides interpretation of statistical results
• Consider practical significance alongside statistical significance in decision-making
• Examples of application:
  • T-test with t-statistic = 2.5, degrees of freedom = 20, two-tailed test: Critical value at α = 0.05 is approximately ±2.086 If |t-statistic| > critical value, reject null hypothesis
  • Z-test with z-score = 1.96, two-tailed test: P-value = 0.05, equal to commonly used significance level Borderline case for rejecting null hypothesis