GOP Candidate YouTube Stats - 06/11/07

Well now, we've got a new contender in the "GOP Frontrunner" analysis of the GOP candidate internal stats at YouTube.com: Mike Huckabee. Seems this secord-tier candidate is doing something right at the popular video sharing site to give frontrunner Giuliani a run for his money. Take a look at the charts:

These represent the cumulative stats for each candidate across all of the videos they've uploaded to their official YouTube page, first looking at just the cumulative number of views (inspired by TechPresident.com's TubeMogul charts), followed by the cumulative number of comments and ratings for all videos per candidate with the number of videos, channel views and subscribers to each candidate. (Click the charts to view fullsize)



Once again, we'll throw away the outlier video data (see below for outlier explanation) because these reflect unusually heavy traffic resulting in lopsided averages that tend to skew the loyalty depth for each candidate, and that's what we're trying to gage: candidate loyalty and support. For example, take a look at how the single outlier video impacts the total views for each candidate's videos:


Once we subtract the outlier data from the cumulative totals, we get a better view of the regular support each candidate has achieved at YouTube.



That's all for this week's data. I want to take a minute just to go back and look at the growth over the last month, using the charts I created with the data that was current for each interval. I'll convert these to graphs later this evening.

Update - here are the graphs showing growth in user activity for each candidate since May 22. Some of this gets into minutiae that may not be very useful, but until that becomes apparent, here they are.

Channel Views	Cumulative Ratings	Cumulative Comments
Cumulative Views	Subscribers	Ratings Averages
Comments Averages	Views Averages

It appears all of the candidates are enjoying steady growth in their loyal support base, except Giuliani. His numbers actually seem to be falling, but it's too early to tell what might be causing this. I suspect the Giuliani campaign isn't putting as much effort into internet campaigning as the other candidates seem to be doing. If that's true, will it hurt him in the campaign overall? The internet community is proving itself to be a significant influence on the success of each candidate, and it's becoming a common rule of the political game to exploit the internet for its maximum benefit. Is the Giuliani campaign slacking off or are his supporters souring on his message? Too soon to tell.

Previous posts:
June 2, 2007: GOP Candidate YouTube Stats, Charts (Followup)
May 22, 2007: About those Youtube Stats...


Update II - June 12, 2007
Several people have asked how I obtained the data and how the outliers were determined so here is what I did.

I signed up for a developer's account at Youtube, giving me access to tons of xml-formatted data that can be downloaded into any xml-compatible application. I then downloaded the user stats for each candidate into Excel to manage the data and generate charts.

To determine how much of a following each candidate has, you have to look at his averages instead of his cumulatives. When one video receives an abnormal amount of traffic, the averages you get for each video, including the one with the abnormal traffic, are not truly representative of the activity directed toward that candidate. For example, if you look at this Excel screen grab for the Youtube data for Duncan Hunter, the negative impact of outliers becomes very apparent.

He has 17 videos in his channel, his cumulative video views equal 353,271, and one of these videos has received 293,862 views. When you include this outlier data in the averages, Hunter's video view average is 20,781 per video. However, given that none of his remaining 16 videos have received more than 8,522 views, this 20,781 average is a false representation of the breadth of his supporters. Therefore this outlier data is set aside in the calculation of user averages to get a more accurate representation of his support.

As I explained in my first post, most of the candidates in frontrunner status at Youtube (as originally defined by TechPresident.com) had one video that blew all the others away for reasons that are impossible to attribute solely to user devotion. It is likely these individual videos and their unusually high traffic numbers were the result of YouTube's spotlight effort, or the possibility that it was embedded on a high-traffic website for automatic viewing each time the page was loaded, without regard to viewer interest in the video content itself. There is no way to know for certain whether these high traffic numbers were from genuine candidate support, from general public exposure which would include views from both friend and foe alike, or from the efforts of page-refresh scammers who use common tricks of the web to drive up the number of views on that video. To measure a candidate's genuine support, which is my primary concern, you have to look at how his other videos are treated and whether his information is being sought and viewed intentionally. For that reason, all of the candidates' averages for my analysis were calculated with the highest-traffic video data removed from the totals.

I hope this explains sufficiently how I arrived at the conclusions I did in these analyses.

---