My dep variable is investment (% of GDP) and my independent variables are saving (% of GDP), Terror (which is an index I have created, where a value of 0 is the lowest level of terrorism and 1 is the worst level of terrorism) and an interaction term between saving and terror.
I am struggling to know how to interpret my coefficients
save --> 0.6308
terror ---> 33.549
interaction ---> -1.467
I have also run the same regression, but using number of attacks as the terror variable. In this case I get these coefficients:
save ---> 0.6278
terror ---> 0.059
interaction --> - 0.0026
If anyone can help me interpret these results I would be very grateful!!
Your dummy is there to show whether if as a result of intensity of terrorist attack there is a level shift in investment, that is your intercept drops down or not (assuming you only have intercept dummy). The rest are % changes so to see % change in DV as a result of 1% in IV variable. Take a look at the "sign and size" of coefficient and the p value and if it is significant then it it tells the change is significant.
Anna Y Hillingdon
First of all, thanks for your reply!
In the case of using the index for the measure of terrorism, is it a dummy? Because it's not that it can only take on discrete values- it can take on the whole range of values in the 0 to 1 range. Therefore, when I talk about it in my paper do you suppose I should evaluate the subsequent coefficient on saving for the year with the lowest index measure for terrorism, and compare it to the subsequent coefficient on saving for the year with the highest terrorism index value?
Or, for the analysis, should I go down the line of: A 10% point increase in terrorism index (i.e. a 0.1 increase in the index value) decreases the coefficient on saving by 0.147 [would that mean 14.7% points or just 14.7%]?
Thanks in advance for any help!
In which case it is not a dummy.
if you have more than one indepndent variable then you are looking at the partial effect.
I don’t know about your index so my answer is general.
what happens to investment if intensity of your index increases by 1% ? beta should tell you this.
take a look at Gujarati and porter (2009) chapter 6.
Anna Y Hillingdon The coefficient on my index (index Bi in screenshot) is 37.88.
My index has been created such that 0 = the least amount of terrorism in the period and 1 = the most amount of terrorism in the period.
I think this means that a 0.1 increase (or 10% point increase) in severity of terrorism leads to a 3.788% point increase in the investment rate?
I presume your independent variable is a proportion so that the beta (coefficient) implies an elasticity of 0.3788. Just to make sure multiply your index by 100 re-estimate the model so it is easier than dividing.
Anna Y Hillingdon Okay brilliant, thank you Anna.
I've attached the new output and yes the coefficient has fallen to 0.3788.
Just to check then, as this is now an elasticity the interpretation would be that a 1% increase in the Terror index leads to a 0.39% increase in the investment rate (i.e. investment as a percentage of GDP).
Also, a 1% increase in the Terror index decreases the coefficient of saving (as a percentage of GDP) by 0.016%? [should this be 0.016 percentage points?]
Anna Y Hillingdon You've been such a help, thank you so much I really really appreciate it!!
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering