Lecture Notes: Testing Causal Hypotheses

Introduction

• Focus: Testing causal hypotheses and establishing relationships between variables.
• Previous Discussion: Use of Excel for descriptive statistics (mean, median, mode, standard deviation).
• Example Research Questions:
  • Impact of gender on political affiliation.
  • Level of violent protests in different political regimes (democracies vs. non-democracies).
  • Student-to-teacher ratio variations in different types of schools.

Basics of Hypothesis Testing

• Main Idea: Comparison of groups.
• Steps:
  1. Divide sample based on values of the independent variable.
  2. Compare dependent variable values (e.g., income differences between genders).
• Conclusion Logic: If dependent variable means are similar, no relationship exists; otherwise, a relationship is present.

Examples of Hypotheses

• Gender influence on political affiliation.
• Level of violent protest comparison between democracies and authoritarian regimes.

Testing Procedures

• Comparison of the Means:
  • Used when the dependent variable is interval and the independent variable is nominal/ordinal.
  • Steps:
    • Separate groups based on independent variable.
    • Calculate mean of the dependent variable for each group.
  • Applications: Drug effectiveness studies, health effects between smokers and non-smokers.

Using Excel for Hypothesis Testing

• Excel Function: =AVERAGEIFS
  • Allows calculation of means for dependent variables based on independent variable values.
  • Structure:
    1. Start with an equal sign and AVERAGEIFS.
    2. Enter range of cells for dependent variable.
    3. Enter range and specific values of independent variable.
• Example: Calculating average income for males and females.

Practical Example

• Research Question: Is the level of corruption lower in democracies compared to non-democracies?
• Variables:
  • free_corrupt: Ranges from 0 (very corrupt) to 100 (no corruption).
  • democracy_regime: Values 1 (democracy) and 0 (authoritarian).
• Steps:
  1. Use =AVERAGEIFS to calculate the average corruption level for authoritarian states and democracies.
  2. Compare the two averages.
  3. Conclude if a significant difference exists (e.g., 16 percentage points difference suggests democracies are less corrupt).

Conclusion

• Significance: Large differences indicate a relationship (e.g., democracies vs. non-democracies in terms of corruption).
• Further Resources: Examples and tutorials available on YouTube for using Excel in statistical analysis.

Contact

• Instructor Availability: Contact for further questions and additional examples.