Deck 5 - Dummies and Functional Forms

Question 1

how many dummies for a two-category attribute

Answer 1

one (avoid the dummy variable trap).

Question 2

interpreting a dummy in levels

Answer 2

β for dummy = level difference vs the omitted base group.

Question 3

log-level model (log y on x)

Answer 3

1-unit increase in x changes y by ≈100×β% (exact: 100(e^β−1)%).

Question 4

Level-log model (y on log x)

Answer 4

1% increase in x changes y by β/100 units.

Question 5

Log-log model (elasticity form)

Answer 5

% increase in x changes y by β% (β is elasticity).

Question 6

Dummy in log-level model

Answer 6

Exact % effect = 100×(e^β−1).

Question 7

Quadratic model marginal effect

Answer 7

dE[y]/dx=β1+2β2 x; turning point at −β1/(2β2).

Question 8

When to log a variable?

Answer 8

Positive, right-skewed variables where proportional changes are meaningful.