Feedback

[bensoltoff]

• You have a very nice literature review. Succinctly summarizes the main theoretical approaches to explain how targeted killings may effect militant organizations, and sets you up nicely for your own hypothesis tests
• Keep in mind really you have just a single hypothesis - how fatalities from covert air strikes effect the intervals between attacks
• Nice descriptive statistics graphs

  The parameters were set arbitrarily as follows:

  • Why were they set arbitrarily? Could you not have tested to determine the optimal values? Do these come from theory/past research?

• You misinterpret your regression results

  A 1% increase in the cumulative effect measure is associated with a 0.07% decrease in the number of days until the next attack.

  • This is not correct because you log transformed both the independent and dependent variable. It should read A 1 unit increase in the logged cumulative effect measure is associated with a 0.07 unit decrease in the logged number of days until the next attack.

• A bit light on the computational elements, though interesting approach to the OLS model. Unfortunately also impacted by the clearly incorrect interpretation of the OLS results. And we specifically discussed interpreting logged coefficients last term
• I don't particularly care that the results are not what you expected, but the model is somewhat weak in design. You don't account for any other plausible explanations for the interval length between terrorist strikes that could confound the relationship between covert airstrike fatalities and interval length

[kimswchi]

I interpreted the regression results in terms of elasticity. I used the coefficient interpretation guide Ging Cee sent us (the Log-Log section): http://isites.harvard.edu/fs/docs/icb.topic1210732.files/Coefficient%20Interpretation.pdf

[bensoltoff]

Ahh, okay that makes more sense