Exam Technique (what to answer to each Q)

Question 1

What large question is the author trying to answer? (5 Marks)

Answer 1

Simple one or two sentence answer of the hypothesis

Question 2

Briefly explain the authors research design? (10 Marks)

Answer 2

• Sources of data
• What type of regression is being done?
• Type of data (panel, cross sectional, time seriess) with explanation of sources
• What are the main controls

Question 3

hat is the main, take home conclusion from the table? (15 Marks)

Answer 3

• correct interpretation of coefficients
• discussion of statistical significance
• magnitude of coefficients; historical significance
• explanatory power of the model; r squared. Omitted variables?

How does new controls or different subsamples change the above

Question 4

What do the regression coefficients mean, are they statistically significant, and what is the explanatory power of the model?

How do these factors change when adding new controls or limiting the analysis to a subset of the data? You do not need interpret every coefficient. Just the ones that are important for the main conclusions you draw from the table. (30 points)

Answer 4

correct interpretation of coefficients
* discussion of statistical significance
* magnitude of coefficients; historical significance
* explanatory power of the model; r squared. Omitted variables?

How does new controls or different subsamples change the above

Question 5

Are there any potential drawbacks with the type of data used? (20 Marks)

Answer 5

• See disadvantages of data and holess in it with regardds to conlusion
• Focus on data type and potential bias or representatives.
  Analyse the biases they could arise in detail
  Understand how those biases may change over time or by region depending on data types

(add more

Question 6

What are the likely pitfalls with the authors method of choice? (20 Marks)

Answer 6

Examine the data as well as the author’s research design to understand the strength of his assumptions

Look for Collider biases and potential confounders

Look for potential omitted variables that may be problematic for the analysis; think about correlations or relationships that aren’t accounted for that distort the pattern

Also look to see if used dummies only, if so the author is assuming similar
Revalues across all covariates.

Look in notes if something more complex (interactions, IV, Diff in Diff)

Look for problems with regression and if there may be types of selection biases