I recorded the web traffic from each site from January 19 to February 17.  Google Analytics records the number of unique users and the number of page views from each day.   The correlation coefficients for the web traffic estimates are listed in the table above.  The google analytics estimates of the users and page views were most highly correlated with each other at 0.886 (on a scale from -1 to +1).  Google Adsense estimates were most highly correlated with Google Analytics page views .  Blogger’s built in traffic counter measures page views and is most correlated with Adsense.  Statcounter is a site not affiliated with Google and not surprisingly is not strongly correlated with any of their traffic estimates with Google Analytics pageviews being strongest at 0.101.

The above graph shows how Statcounter (in black) compares to Google Analytics and Adsense.  There was no similar spike in traffic on that day.  There were spikes in traffic on both statcounter and google analytics on other days for page views.  The top five most visited pages are displayed for each site below.  No pages appeared in all three sites.  Some pages appear in two out of the three sites such as Best Nonfiction Poll Update and The Civil War in a Larger International Historical Context.  Statcounter only provides the top pages for the last 12 days for free.  

I don’t which of these sites are right.  I’m not sure which is the gold standard.  There is certainly some noise in the signal.  Because Google owns Blogger and a host of other analytics products it’s good to have an outside site to validate the numbers.  Google determines ad revenue so Adsense has the final say.

Cambridge Analytica was Behind the Results I Found on Facebook and the 2016 Election

Podcamp Pittsburgh XI