7
$\begingroup$

I am interested to examine the effect on consumption on marijuana with the implementation of tax. I will be examining it on 51 states in US including DC, over 6 years. The other controlled variables would be unemployment rate and education. Tax would be a dummy variable, 1 will take form when the state taxed on marijuana while 0 will take form when marijuana is not taxed. However, my advisor suggest that the consumption and taxation would have reverse causation. Can I know what is the best method in addressing it? I would also need to include time fixed effect in my model.

$\endgroup$
2
  • $\begingroup$ In the version you presented to your advisor, which variable is the dependent and which is the independent? $\endgroup$ Dec 14, 2020 at 15:08
  • $\begingroup$ @Penguin_Knight the dependent would be consumption of marijuana while the independent would be tax $\endgroup$
    – Emily Jane
    Dec 14, 2020 at 15:21

1 Answer 1

7
$\begingroup$

In this case it'd be helpful to know more about i) the nature of the consumption data, what are included and how were they collected? ii) if any states change their taxation status across the 6 years, or do all of them show either all six years of 1, or all six years of 0?

If you do have some switching over, then perhaps interrupted time series regression could be one of the options. It uses each state as its own control, and examines if the regression slope changes before and after the implementation. (You may need to center the year at implementation for each state, so that year 0 is the year of implementation.)

The main concern I'm seeing is that right now: there are only about 20 states that have legalized marijuana. And in the last 6 years, even if you have some states switching over, the sample size is likely very low. It may be better to first identify which states, and then download the corresponding years of BRFSS data from CDC, and create a bigger individual level data.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.