pol222 notes without flashcards

Key Concepts:

  1. Types of Analysis:

    • Qualitative Analysis:

      • In-depth study of a few cases (historical accounts, interviews).

    • Quantitative Analysis:

      • Use statistical methods to study large datasets.

      • Benefits: Identifying patterns, testing theories on many cases.

  2. Why Quantitative Methods?

    • Build generalizable knowledge about politics.

    • Avoid selection bias by examining diverse datasets.

    • Example: Understanding why some countries democratize using global data rather than selective case studies.

  3. Methodology Overview:

    • Theory and Hypothesis:

      • Develop causal theories and testable hypotheses.

    • Empirical Analysis:

      • Collect and analyze data to validate hypotheses.

 

Week 2: How Do We Study Politics Quantitatively?

Applications of Quantitative Methods:

  1. Forecasting:

    • Predict election outcomes, economic trends.

  2. Description:

    • Use surveys to measure public opinion, ideology, and democracy.

  3. Causal Questions:

    • Evaluate policies and their outcomes using causal theories.

Scientific Method for Causal Questions:

  1. Theory:

    • Framework explaining causal relationships (e.g., economic development leads to democratization).

  2. Hypothesis:

    • A testable prediction based on theory.

  3. Empirical Analysis:

    • Use real-world data to confirm or refute hypotheses.

 

Week 3: Evaluating Causal Relationships – Part 1

Challenges in Causal Inference:

  1. Confounding Variables:

    • Factors (Z) that influence both the dependent (Y) and independent (X) variables.

  2. Reverse Causality:

    • Difficulty in determining whether X causes Y or vice versa.

Key Terms:

  • Omitted Variable Bias:

    • Excluding confounders leads to misleading conclusions.

  • Spurious Relationships:

    • False associations that disappear when confounders are controlled.

Examples:

  • Misinterpreting educational attainment as the sole cause of political participation without accounting for socioeconomic background.

 

Week 4: Evaluating Causal Relationships – Part 2【16†source**

Biases from Omitted Confounders:

  1. Overestimation:

    • Effect appears stronger than it is.

  2. Underestimation:

    • Effect appears weaker than it is.

  3. Sign Reversal:

    • Effect appears in the opposite direction due to confounding bias.

Approach:

  • Account for confounders using statistical controls or advanced modeling.

 

Week 7: Experimental Research Basics

What Makes Experiments Powerful?

  1. Random Assignment:

    • Balances confounders across groups.

  2. Controlled Environment:

    • Isolates the effect of independent variables on outcomes.

Challenges in Political Science:

  • Difficulty replicating real-world complexity in experiments.

Example:

  • Testing the impact of negative political ads on voter turnout:

    • Control group sees neutral ads, treatment group sees negative ads.

    • Compare voting intentions.

 

Week 8: Validity in Experimental Research

Types of Validity:

  1. Internal Validity:

    • Confidence in causal relationships.

    • Enhanced by random assignment.

  2. External Validity:

    • Generalizability to real-world scenarios.

    • Often limited in lab settings but stronger in field experiments.

Example Studies:

  1. Garramone et al. (1990):

    • Lab study on political ads with controlled stimuli.

    • Strong internal but weak external validity.

  2. Ansolabehere et al. (1994):

    • Field study with real election ads.

    • Better external validity.

 

Week 9: Survey Experiments & Discussion【19†source**

Survey Methods:

  • Representative Sampling:

    • Randomly selected participants allow generalization to the population.

  • Non-Representative Sampling:

    • Convenience sampling is less generalizable.

Survey Experiments:

  • Treat survey questions as experimental stimuli.

  • Example: Testing support for military action under conscription vs. voluntary service.

Ethical Considerations:

  • Ensure participants are informed and not harmed.

 

Week 10: Observational Research【20†source**

Why Observational Studies?

  • When random assignment is infeasible, use real-world data to examine X → Y relationships.

Approach:

  1. Conditioning:

    • Compare outcomes (Y) under different independent variable values (X).

  2. Controlling for Z:

    • Adjust for confounders statistically to reveal true effects.

Examples:

  • Economic voting in Canada:

    • Perceptions of the economy’s health predict voter preferences.

 

Week 11: Linear Regression Analysis【21†source**

Purpose of Regression:

  • Model relationships between variables.

  • Predict outcomes (Y) based on independent variables (X, Z).

Key Concepts:

  1. Simple Regression:

    • Examines one independent variable.

  2. Multiple Regression:

    • Accounts for multiple confounders.

Interpretation:

  • Coefficients indicate the impact of X on Y.

  • Example:

    • Improving economic perception increases support for the incumbent party by a measurable amount.

 

Week 12: Validity of Observational Research & Natural Experiments

Internal and External Validity:

  • Internal validity may be weaker than experiments.

  • External validity often stronger than lab experiments due to real-world context.

Natural Experiments:

  • Exploit naturally occurring variations resembling random assignment.

  • Example:

    • Study effects of policy changes or natural disasters.

Key Focus in Tutorials:

  • Analyzing case studies for observational and natural experiment designs.

  • Identifying and mitigating potential biases.

Here’s a clear breakdown of these types of data collection:

1. Cross-Sectional Data:

• Definition: Data collected at a single point in time or over a very short period.

• Characteristics: Provides a “snapshot” of a population, group, or phenomenon.

• Example: A survey of 1,000 people’s voting preferences conducted in December 2024.

2. Time Series Data:

• Definition: Data collected over time at regular intervals (e.g., daily, monthly, yearly).

• Characteristics: Focuses on trends, patterns, or changes over time.

• Example: Monthly unemployment rates in the U.S. from 2000 to 2023.

3. Panel Data:

• Definition: Data collected over time from the same subjects or entities.

• Characteristics: Combines the features of cross-sectional and time series data; useful for tracking changes at the individual level.

• Example: Tracking the income levels of 500 households annually for 10 years.

4. Time Series Cross-Sectional (TSCS) Data:

• Definition: Data collected over time across different entities (not necessarily the same ones every time).

• Characteristics: Similar to panel data, but the entities observed might differ across time periods.

• Example: Measuring GDP and inflation rates for different countries over 20 years, where the countries studied might vary over time.

Key Difference:

• Cross-sectional is a single time snapshot.

• Time series is continuous over time for one entity.

• Panel tracks the same entities over time.

• TSCS involves multiple entities over time but doesn’t require tracking the same entities throughout.