Online video ad measurement for political science research

With over 1 billion users and over 4 billion video views a day, YouTube is the most popular online video sharing website. Over a million advertisers use the Google ad platform which provides the ability to target specific groups of users. Political campaigns have used YouTube to reach voters. However, political science literature does not have information about how online video ads are targeted to voters via YouTube. This research made two contributions. (1) We developed “CyAds Meter,” a unique web bot platform that collects information about video and image ads delivered on YouTube. Our bot system adheres to YouTube's terms of service and avoids financial cost to online advertisers. Our two-tailed Wilcoxon-Mann-Whitney test verified that the bot and the human with the same online profile see the same number of ads on average with 5% significant level. In other words, our bot does not significantly miss ads nor see more ads than the human does. (2) We used CyAds Meter to provide evidence for political science scholars that the numbers of ads aired in the battleground and non-battleground states prior to the 2016 presidential election is statistically significant different at a 5% significant level. CyAds Meter is a potentially useful for political science research to study micro-targeting tactics of online ad election campaigns.