I'm reading a paper that tries to estimate the following model: $$y = \alpha x+\beta z+\gamma xz+\epsilon$$ where x,z are dummy variables and y isn't. The estimated coefficient of $\alpha$ is positive (0.02), and the sum of $\alpha+\gamma$ is negative (-0.01,insignificant). The writer examines the marginal effect of x, conditional on z and finds a positive (0.01, with significant of 90%). My question is what is the difference in interpretation of marginal effect VS. $\alpha+\gamma$?

The paper is "Terrorism and Voting: The Effect of Rocket Threat on Voting in Israeli Elections" by Anna Getmansky and Thomas Zeitzoff (link: https://www.zeitzoff.com/uploads/2/2/4/1/22413724/zeitzoff_getmank_rockets_main.pdf). The marginal analysis is on page 11.

  • $\begingroup$ can you include a link to the paper you are referencing? $\endgroup$
    – EconJohn
    Dec 31 '17 at 16:22
  • 1
    $\begingroup$ I added a link. $\endgroup$
    – Neta_1990
    Dec 31 '17 at 17:02
  • 1
    $\begingroup$ Wow, I didn't know they did quantitative work on this. Seems like a little bit of a wild topic to be researching. $\endgroup$
    – EconJohn
    Dec 31 '17 at 17:20
  • $\begingroup$ @Neta_1990 I edited the title of your question. Please edit the title again if you feel my edit was inaccurate. $\endgroup$ Dec 31 '17 at 19:22

The paper computes the value of your coefficient $\gamma$ on its own with comparison to $\alpha +\gamma$1.

The interpretation in this context, would be that there is a limit of the returns to RightShare given the presence of observations existing contemporaneously with RightPM and InRange.

1. this of course assumes that $z=1$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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