Of the 3,012 records in this dataset, 217 were targeted at users in specific locations. I imagined at the outset that the location-based ads would be primarily focused on manipulating public opinion in swing states like Pennsylvania, Ohio, and Florida; however that does not appear to be the case here. As this map illustrates, the majority of IRA Facebook ads that did target users in specific locations clustered around cities that featured recent racial tension—namely Ferguson, Missouri; St. Louis, Missouri; Milwaukee, WI; Baltimore, Maryland; and Atlanta, Georgia. As many have pointed out, and as this map indicates, racial tensions proved to be a crucial instrument for the IRA’s campaign. But if there were so many cities where racial tensions were being inflamed, I was puzzled as to why location-targeted ads comprised such a small percentage of the total ads. I wondered whether I could calculate how effective these location-based ads were in generating impressions and clicks, so I created the table below.
| ENTIRE DATASET | LOCATION-BASED ADS | |
|---|---|---|
| TOTAL RECORDS | 3012 | 343 |
| TOTAL COST | $88,512 | $17,157 |
| TOTAL IMPRESSIONS | 37,658,631 | 2,677,548 |
| TOTAL CLICKS | 3,414,792 | 169,252 |
The location-based ads accounted for 19.38% of the cost of the entire collection of 3,012 ads, but they only accounted for 7.11% of the total impressions and 4.96% of the total clicks. So it’s clear that the location-based ads were not the most cost-effective method employed by the IRA. Perhaps they realized that in a globalized world with the sheer number of users on social media, it’s still not very difficult to spread disinformation even when it’s not targeted at a specific location. Perhaps they also realized that through effective use of tagging and racial affect, they could more efficiently spread their content over wider swathes of the country.