Research Methods

​

Roman Sheremeta, Ph.D.

Professor, Weatherhead School of Management

Case Western Reserve University

1

Part 1: Inferring Causality

Part 2: Theoretical Method

Part 3: Experimental Method

Part 4: Econometric Method

​

​

​

​

​

​

​

• .

2

Part 1: �Inferring Causality

3

Correlation versus causation�

• Correlation: To say that a variable X is correlated (or associated) with a variable Y means that the variables “move together” in the data.

​

​

​

• Correlation helps us to predict an outcome.
• If X and Y are correlated (or associated), then knowing the value of X helps you predict the value of Y (and vice versa).

​

• Correlation does not imply causation.
• Just because two things occur together does not mean that one caused the other, even if it seems to make sense.

4

Example: ice cream�

• Fact 1: Increased ice cream sales is associated with increased drownings.
• Conclusion 1: Consumption of ice cream leads to more drownings.
• Conclusion 2: There is an omitted variable – temperature. A hot summer day will boost ice cream sales and swimming. And more swimming will result in more drownings.

​

​

​

​

​

​

​

5

Example: education�

• Fact 2: For each year of education a person in the United States earns approximately $4,000 more per year.
• Conclusion 1: Education leads to higher earnings.
• Conclusion 2: Smarter people choose to go to the university, and smarter people will be able to find better jobs not necessarily because of their education but simply because they are smarter.

​

​

​

​

​

​

6

Inferring causality�

• Causation: To say that a variable X is causally related to a variable Y means that changing the value of X in reality would lead to a change in the value of Y.

​

• How to infer causality?
• Y may be a function of many factors other than X.
• We want to measure any changes in Y that are directly attributable to a change in X, not to these other factors.
• To do so, we have to think about the counterfactual: What would have happened otherwise if X did not occur?

​

• What is the counterfactual for each of these two examples?
• Does getting PhD (X) affect your job prospects (Y)?
• Do industrial emissions (X) cause the temperature of the planet (Y) to rise?

​

7

Research methods�

• Economic research is directed at examining causal relationships, not simple correlations

​

• Methods of economic research:
1. Theoretical method
2. Experimental method
3. Econometric method

​

​

​

​

​

8

Part 2: �Theoretical Method

9

Theoretical model�

• Model: a theoretical description of the relationship between two or more variables.
• Meteorologists use models to predict weather conditions.
• Medical researchers use models to describe and predict the effect of medications on diseases.
• Astronomers use models to describe and predict the movement of planets.
• Economists use economic models to explain how managers and other economic agents make decisions and to explain the resulting market outcomes.

​

​

​

​

​

​

10

How to build a model?�

• Theoretical method (Paul Samuelson, John Nash):
1. Assumptions (axioms).
2. Theoretical analysis (general equilibrium, game theory, agent-based modeling, linear and non-linear programming, dynamic optimization, simulations).
3. Predictions (comparative statics analysis).

​

​

​

​

​

​

​

​

11

Part 3: �Experimental Method

12

Experimental method�

• Experimental method: Learning from observing behavior in an environment created or modified by the researcher for that very purpose.
• Controlled experiment with random assignment is the Gold Standard for inferring a causal relationship between X and Y.
• The experiment allows the researches to change one variable at a time (X) and see how this variable impacts the outcome variable (Y), while holding all other variables constant.

​

​

​

​

​

​

​

13

How to run an experiment?�

• Experimental method (Vernon Smith, Richard Thaler, John List):
1. Hypothesis (based on theory).
2. Experimental design (randomization, treatments, procedures, laboratory or field).
3. Data analysis (non-parametric and parametric tests, regressions).

​

​

​

​

​

​

​

14

Part 4: �Econometric Method

15

Observational study�

• Observational Study: the researcher analyzes data from a situation over which he or she has had no control.
• Example: people who smoke more (X) have a higher likelihood of getting a lung cancer (Y). Ronald Fisher has famously debated this issue in 1950s.

16

• What are the common problems with observational studies?
• Measurement error.
• Omitted variables (ice cream example).
• Simultaneity (more police → less crime, more crime → more police).
• Selection (individuals are not assigned randomly into treatments).

Econometrics�

• Econometrics: provides statistical techniques that allow us to try to isolate the “causal” effect of X on Y when we have observational data.

​

• Econometric method (Steve Levitt, James Heckman):
1. Hypothesis (based on theory).
2. Data collection.
3. Econometric/statistical analysis (MV, IV, DID, RDD).

​

​

​

​

​

17

References�

• Varian, H.R. (2016). Causal inference in economics and marketing. Proceedings of the National Academy of Sciences, 113, 7310-7315.

​

18