Which of the following are correct statements regarding Difference in Differences (DID) methods? (choose all correct answers).

A. DiD models can lead to inaccurate answers if there are pre-trends in either the treatment or the control groups.

B. Did models can be affected by reverse causality, where the outcome variable influences the treatment, not the other way around.

C. Omitted Variable Bias can affect DiD models. Here an unaccounted for "third factor" affects the levels of both the treatment and control measures.

D. An instrumental variable is a potential solution for reverse causality and omitted variable bias. The two conditions for an instrumental variable are that (1) it only affects the outcome variable through the independent variable of interest; and (2) it is not correlated with any omitted variables.

E. The DiD estimator takes the ratio of the post and pre levels in the control group (e.g. post/pre) and divides by the ratio of the post and pre levels of the outcome in the treated group (e.g. post/pre).

F. DiD models can lead to inaccurate answers if there are pre-trends in the treatment group. Pre-trends in the control group will not affect the DiD estimates.

G. DiD models examine the effect of some change on an outcome. They compare two "treated groups that were both exposed to a policy change to examine endogeneity and heterskedasticy concerns.

H. The DiD estimator takes the difference between the post and pre levels of the outcome in the treated group (e.g. post-pre) and subtracts the difference between the post and pre levels of the outcome in the control group (e.g. post-pre).

I. DiD models examine the effect of some change on an outcome. They use a "treated group" that was exposed to the change and a "control group" that was not exposed to the change.