One of the 24 hour news networks recently ran this quiz question: “Which country spends the most time on Facebook?” They listed a number of possible answers including India, the U.S., New Zealand, and Singapore. While trying to figure out the answer, I started to wonder what they measured. Is it the total time for the country? This might favor India with so many people. Or, is it time per user? And, if it’s per user, what is the time frame, one session or a specific time period? One question led to a lot of questions!
Researchers, Six Sigma practitioners, and anyone else who uses statistics to make good decisions must figure out how to collect and measure data that will answer the question at hand. Sometimes statistical analysis can be difficult when you have data limitations or a complex topic to measure. It turns out this Facebook study was conducted by Techlightenment, an Experian company. I suspect that they ran into both of these issues. In this case, how do you measure and then report statistics for something as complex as how users access a service like Facebook?
The “correct” answer is Singapore. It turns out that the news network, as well as many online news articles, determined this because Singapore had the highest average Facebook session length. Each time a Singaporean logged on to Facebook they stayed for over 38 minutes, while U.S. citizens were ranked fifth at about 20 minutes per session.
That’s interesting, but determining "which country uses Facebook the most" is a much more complex question. For one thing, there’s an obvious follow-up question regarding the length of Facebook sessions: what about the frequency of sessions? It’s possible that while Singaporeans have a longer average session length, the citizens of another country may log on to Facebook much more frequently.
There’s also the question of how popular Facebook is in each of the countries. While the session length is long in Singapore, perhaps Facebook isn’t that popular in Singapore overall?
Experian’s full report does offer the answers to some of these questions. For instance, the relative popularity of social networking and Facebook varies by country. It turns out that Brazil has the largest network share (18.9%) of visits going to social sites. However, Brazil’s most popular social site is Google’s Orkut, and Brazilian Facebook sessions are among the shortest. Singapore has the second highest share of Internet traffic going to social networks, and 18% of social site users jump directly from one social network to another. In India, Facebook sessions are short, but that's where Facebook’s share of Internet traffic is growing at the fastest rate. In the U.S., Facebook accounts for 91% of network traffic to social networking sites this August.
So, we can get a sense of some of the variables at play and there are many different ways to answer the question about Facebook usage, particularly when you look at the larger context of overall social network usage and changes over time. However, some crucial questions remain unanswered, and I strongly suspect that this is due to limitations in the data that the researchers can collect. Experian reports some large-scale changes over time, such as the growth in traffic over a year. For example, Facebook’s share of network traffic in the U.S. increased by 5% over the past year. However, Experian’s data does not seem to be able to track specific users over time. If this is true, this limitation means that they will be unable to determine that important issue about the frequency of sessions for Facebook users.
Not tracking specific Facebook users over time is probably good for privacy. However, this seemingly simple news quiz underlines the importance of understanding how limitations in your data can affect the conclusions that you can draw from it during data analysis.
(And if you're a Facebook user yourself, be sure to visit and "like" the Minitab Facebook page!)