Saturday, July 28, 2012

The non-science behind Bluefin Labs' Opening Ceremony social media analysis #nbcfail

There are many social media studies that are conducted that can tell you all sorts of wonderful things. But before you trust any study, you should take a careful look at the methodology behind the study.

Case in point - Bluefin Labs has posted an item entitled How Social Was the Olympics Opening Ceremony? It quotes an impressive-sounding number, but there is a very large flaw in their methodology. See if you can spot it.

As expected, the Opening Ceremony of the 2012 London Olympics was huge in social TV. How huge?

94.2% of all social TV comments in primetime (7pm to 12midnight ET) last night were about the Olympics Opening Ceremony.

The Bluefin post then displays a really impressive graphic that states that this social media spike occurred when Queen Elizabeth parachuted out of the helicopter, this spike occurred when Team USA entered, and so forth.

There are two problems with this analysis.

If you've been around me for any length of time, you already know what one of them is. Bluefin defined prime time as occurring between 7:00 pm and 12:00 midnight Eastern time. Perhaps that was true for some people, but it wasn't true for the western part of the United States. In California, NBC didn't begin showing its Olympic coverage until 7:30 pm Pacific time - or 10:30 pm Eastern time. NBC showed Paul McCartney's rendition of "Hey Jude" at about 3:00 am Eastern time.

This causes a methodology problem. If you look at Bluefin's chart, you'll see several blips between 11:00 pm and 12:00 midnight Eastern time. Which of those blips were due to the televised image of David Beckham that was being shown on the East Coast? Alternatively, which of those blips were due to the televised image of Parachuting Lizzy being shown on the West Coast? Unless Bluefin has a methodology to exclude West Coast tweets from its analysis, we'll never know.

But there's a bigger problem - and again, if you've been around me for any length of time, you know what the second problem is. Bluefin was measuring public comments on Twitter and Facebook. But Twitter and Facebook are available in countries outside of the United States. And despite NBC's successful efforts in preventing many people in the United States from watching the Opening Ceremonies at 4:00 pm Eastern time, WHEN THEY ACTUALLY HAPPENED, people - including Americans - were using Facebook, Twitter, and other services to monitor the Opening Ceremony long before Bluefin's study kicked in. For example, I wrote this at 2:19 pm Pacific time, well before NBC Chicago encouraged viewers to "[w]atch NBC's coverage of the Opening Ceremony beginning at 6:30 p.m. CT."

OK, you might say, but Bluefin is concentrating on televised coverage of the event. Well, in most of the world the event WAS televised at 4:00 pm Eastern. Even Canada was showing the event at 4:00 pm Eastern, as I discovered via Google+.

This, of course, makes it even more difficult to say that this blip in Twitter resulted from this event being televised at the time.

Back on January 1, 2008, when hashtags were still relatively new (at least on Twitter), I wrote a post entitled Hashtagging Challenges When Events Occur at Different Times in Different Locations. It described what happened when I was tweeting about KTLA's Rose Parade coverage from a vantage point near the beginning of the parade, while Phil Hodgen was tweeting from a vantage point near the end of the parade. The result was a jumble.

philiphodgen : #roseparade marines get heartfelt applause << (2008-01-01 12:22:58)

philiphodgen : #roseparade boy and girl scouts with state flags << (2008-01-01 12:22:40)

oemperor : #roseparade usc trojans. i only hear this song every week (my alumna coworker has it in her cube). actually i heard it less this year << (2008-01-01 12:22:26)

So were the USC Trojans marching with the Marines and the Scouts? Not exactly.

Now try to cover the tweets, Facebook posts, and other items from the Olympics Opening Ceremony coverage, where you may see the same event seven HOURS earlier or later than someone else. An Englishperson or Canadian might see something at 1:00 pm Pacific time which I wouldn't see until 8:00 pm Pacific time.

How are you going to analyze THOSE tweets? Especially this live tweet from Tim Berners-Lee? (Well, live for those on Twitter; not live for those watching NBC.)

blog comments powered by Disqus