Week 5 : Measurement Error, Simultaneity and Reverse Causality

Question 1

List 3 types of errors that survey data comes with.

Answer 1

• Limited number of options
• Lack of attention
• Coding and transcription errors

Question 2

What are the consequences of random errors in X?

Answer 2

Estimated relationship is weaker than the true effect

Question 3

What are the consequences of random error in Y?

Answer 3

There is no bias

Question 4

What if the errors are not random?

Answer 4

Systematic errors arise when the error itself is correlated with X or Y
- Here we need to think carefully and reason about the possible direction of bias

Question 5

How do we tackle systematic and random errors?

Answer 5

• Collect better quality data, on a large number of observations
• Measure multiple indicators of concepts if possible (especially for systematic errors)
• Refine, check, validate measurement tools
• If aware of a systematic error discuss the implications

Question 6

What is reverse causality?

Answer 6

If Y causes X

Question 7

What is simultaneity?

Answer 7

If Y causes X and X causes Y at the same time.