Journal on Policy and Complex Systems
files are different and the files have different projections , first I transferred all three files to the same projection and then merged them into one shapefile . The second step was to find geospatial data for the four independent variables . Then , I put a value on the independent variables . In other words , based on the existing literature , I picked the data , assigned a numeric value to them , and finally joined these values to the shapefile that I built in the first step . In what follows , I explain the process of allocating numeric values to the independent variables .
I set the value of 0 for provinces located under control of a strong government . 0 means it is hard for the area to cooperate with the Kurdish movement . This group includes all of Iran ’ s provinces , some of Syria ’ s provinces under the control of central government , and some of Iraq ’ s provinces . In addition , I set the value of 1 for regions located under the control of a weak government . For this purpose , other provinces not mentioned above can efficiently cooperate with the Kurdish if they are so inclined .
To evaluate inclination , I consider both being Kurdish and having the same religion . Thus , I placed the value of 0 for non-Sunni Muslim provinces and the value of 1 for Sunni Muslim-populated provinces . I assume that cities populated by Sunni have more inclination to join the Kurdish movement because they perceive some similarity based on religion . In contrast , this inclination declines in cities that are dominated by non-Sunni Muslims .
Similarly , I set the value of 0 for non-Kurdish populated provinces and 1 for Kurdish-populated areas . I assume that somebody should be Kurd or consider her / himself Kurd to join the movement , and vice versa .
Finally , I identified the distance for all provinces to Arbil and classified the distance in five classes . I then set the value of 1 ( farthest ) to 5 ( closest ) based on the province ’ s distance . This reflects the fact that it is much easier for the closest cities to join the movement because they can receive more assistance from Arbil in a shorter time .
At this step of model building , I have four polygon shapefiles : religion , distance , government , and Kurdish populated areas that needs to be transfered to raster shapefiles . After transferring all four to raster shapefiles , I reclassified all four shapefiles to 1-5 since they are supposed to be in the same measure ; a shapefile has 5 classes .
Computational Results and Discussion
I
weighted data based on the percentage of the influence in three different scenarios :
First scenario . In this scenario , I set the same weight ( 25 %) for all four independent variables . I assume here that the central government , distance from Arbil , religion , and being Kurdish have a similar influences .
Second scenario . In this scenario , I assume as this movement is an
30