Lecture 7: Comparison of Means in Sociology 303 Statistics
Introduction
- Instructor: Dr. Alvarez
- Lecture Focus: Comparison of means as a form of bivariate analysis in statistics.
- Goals:
- Understand how to compare averages of different groups.
- Learn to describe and interpret comparative means.
- Test hypotheses using comparative means.
- Introduce and apply dummy variables.
Key Concepts
Comparison of Means
- Used for interval ratio dependent variables (numeric values).
- Independent variables can be nominal or ordinal.
- Dummy variables (dichotomous, coded as 0 or 1) are applicable.
Dummy Variables
- Dichotomous variables (e.g., yes/no, agree/disagree).
- Example: Voting in an election coded as 0 (no) or 1 (yes).
- Calculating mean of a dummy variable gives the percentage of 1s.
Statistical Significance
- Comparison of means entails describing and interpreting the mean for each category.
- Inferential tests are used to determine statistical significance.
Hypothesis Testing
Example 1: Education and Health
- Dependent Variable: Number of days felt healthy and full of energy.
- Independent Variable: Highest degree earned.
- Research Hypothesis: Positive relationship between education and health.
- Null Hypothesis: No relationship between education and health.
Example 2: Lending and Gender
- Dependent Variable: Lending money to friends/family.
- Independent Variable: Gender.
- Description: Analyze using comparison of means and dummy variable.
Example 3: Lending and Race
- Dependent Variable: Lending money to friends/family.
- Independent Variable: Race.
- Analyze racial differences in lending patterns.
Statistical Tests
T-Test
- Used for comparing means of two groups (e.g., males vs. females).
- Steps:
- Determine row (top/bottom) using Levine's test for equality of variances.
- Find p-value to assess statistical significance.
ANOVA and Bonferroni Test
- Used for comparing more than two groups.
- ANOVA provides overall F test result.
- Bonferroni test (or 'Beefaroni') helps specify which group differences are significant.
Application
Example: Christmas Spending
- Research Hypothesis: Households with children spend more on Christmas.
- Null Hypothesis: No difference in spending.
- Analysis: Use t-test or ANOVA based on number of categories compared.
Example: Education and Christmas Spending
- Research Hypothesis: Positive relationship between education and Christmas spending.
- Use ANOVA to analyze differences across multiple educational levels.
Conclusion
- Comparison of means is a crucial technique for analyzing relationships between variables.
- Requires understanding of when to use t-tests vs. ANOVA.
- Describing and interpreting results is key for evaluating research and null hypotheses.
Tips for Success
- Always describe and interpret your statistical findings.
- Use judgment to determine if differences are substantial.
- Understand the importance of statistical significance in testing hypotheses.
This lecture emphasized the importance of applying statistical analysis techniques to real-world data, particularly in sociological research.