Portland Communications’ second How Africa Tweets study has come out right as a massive Twitterstorm rages in South Africa over the Oscar Pistorius trial.
I’ll admit, at the risk of sounding like I’ve been anything less than totally productive these past two weeks, that I have been obsessively following the Oscar Pistorius murder trial. It wasn’t exactly my plan, but as all of my Twitter regulars—Sky News’ Alex Crawford (@AlexCrawfordSky), The Guardian’s David Smith (@SmithinAfrica) and BBC’s Andrew Harding (@BBCAndrewH)—posted trial updates by the minute, the conversation quickly consumed my Twitter feed. By the end of day one, I had my #OscarPistorius page fixed open as I watched the conversation flow. I wouldn’t miss a thing.
So I was more than a little intrigued when I saw that Portland Communications had released new data on How Africa Tweets last week. Portland’s last report, collating and analysing tweets from the last three months of 2011, had been an interesting snapshot of twitter usage on the continent in the face of the Arab Spring.
The newest study found that in the last quarter of 2013, the top five most active cities were in South Africa (Johannesburg, Ekurhuleni and Durban) and Egypt (Cairo and Alexandria). Regionally, Nairobi, Kenya was the most active city in East Africa and the sixth most active on the continent, with 123,078 geo-tagged tweets in Q4 2013, while Accra, Ghana was the most active city in West Africa and the eighth most active on the continent, with 78,575 geo-tagged tweets in Q4 2013.
Geo-tagging means precisely what it sounds like: a tweet is embedded with the latitude/longitude location of the tweeter. If that sounds a bit terrifying, it’s because it is; a lot of people don’t use the geo-tag feature. Seeing as the Portland study was based solely off of geo-tagged tweets, I began to wonder if we weren’t getting the full picture: apparently football was talked about more than Mandela, and Johannesburg’s Orlando Pirates more than any international team. I asked Nanjira Sambuli, a Nairobi-based mathematician and researcher with iHub, her thoughts on the study.
“Now, globally, less than two per cent of tweets are geo-tagged…it follows that few Twitter users in Africa activate this feature. It has privacy implications, for instance, something that frequent users are likely to be aware of,” Sambuli said. “Geo-tagging easily creates a sampling bias issue.”
Not that geo-tags aren’ valuable, she pointed out, it’s just that they’re not used all that often in Africa.
“Geo-tagging is a useful tool, but highly unreliable, especially as a primary or sole metric for analysing tweeting activity in any context. For instance, it was super useful in analysing the Occupy Gezi protests in Turkey last year, because Twitter users consciously geo-tagged their tweets; approx. 90 per cent of 2 million tweets generated within the first 24 hours were geo-tagged. But there’s much more that should go into analysing whether tweeting activity is based in a location at any given time,” she said.
“I must say, I also find it surprising that in the two reports released by Portland Communications, none has acknowledged or disclaimed the shortcomings or limitations of geo-tagging. From an academic perspective, this is something that should accompany every mention or citation of their findings,” Sambuli added.
But when I asked Portland account manager Matt Gould about the geo-tag method, he argued it was in fact the most reliable assessment: “Using geo-location allows us to be certain about each tweet’s true location. Some studies use a tweeter’s stated location (in their bios) but this doesn’t allow any level of certainty since there are no rules or limits about stating one’s location—an individual can note a different city, country or even planet if they want.
“While tracking geo-located tweets does not enable us to capture every single tweet coming out of each city, as many users do not enable locations services, it allows us to develop an indicative picture of overall activity,” he said.
Which brings us to event-inspired tweets, like the drama-filled Oscar Pistorius trial or the harrowing Westgate Mall attack. Could these tweets give us a window into twitter usage in Africa? While the How Africa Tweets study was limited to the last three months of 2013, it did indicate some twitter behavior on incident-based conversation.
“Major local or regional events will definitely have an impact on the level of Twitter activity in any given location,” Gould said. “The death of Nelson Mandela did occur during the period of time we were tracking and we saw a great amount of related activity. The day of Mandela’s death [showed] the largest peak in activity across the continent and his passing was one of the most talked about issues overall, with hashtags like #MandelaMemorial, #Mandela and #RIPNelsonMandela appearing across the continent.”
While the Pistorius trial and the Westgate attack fall outside the scope of Portland’s study, each has proved a window into Twitter activity in Africa: this analysis of the Westgate attack by Sambuli herself found that the incident was first reported by citizens on the scene, mirroring Twitter use in other disasters such as the Boston Marathon attack. iRevolution eventually reported more than 740,000 Westgate Mall tweets during the four-day hostage situation. Tweets were not limited to geo-tags, but rather hashtags (#Westgate, #Westlands, etc.) and other metrics (iRevolution’s full regional breakdown is here).
The Pistorius trial, playing out in real time under the hashtags #OscarPistorius and #OscarTrial, is interesting in another way. The trial—in an unprecedented decision, the first to be televised in South Africa—has generated massive debates on the guilt of Pistorius, the weakness of witnesses, and the ruthlessness of lead defense lawyer Barry Roux. South Africa-based digital media company Fuseware reported that there had been a total of 19,809 mentions of Pistorius from 9,213 South Africans on Twitter by the second day of the trial, and that half of them had been made the first day. The company estimated that there were more than 100,000 conversations on Twitter by the end of the week, and we have about a month of trial to go. Fuseware, it appears, also did not limit themselves to geo-tags.
There is definitely huge interest in the future of Twitter, and social media in general, in Africa. As smartphones become more affordable and internet more readily accessible, Twitter usage on the continent is rising. It’s a matter of time until brands jump in on the buzz—Portland already found that Adidas, Samsung, and Mangum ice cream ranked among the top hashtags at the end of 2013.
In the meantime though, we are still scrambling to find an entirely accurate tool for analysing Twitter use in Africa. However it’s worth noting that this is an international challenge—while we can all agree that everyone uses Twitter a lot, it’s a notoriously unruly social media site, and misleading numbers (or straight-up hoaxes) abound. Gould’s point about geo-tagging is an entirely valid one, as checking someone’s true location when analysing tweets is still a pretty big issue.
Nonetheless, geo-tags don’t give the full picture. “In the future, it would be great if other factors were considered before such findings are put out and cited widely. There’s an appetite for statistics from Africa, and even more so on social media use/number of social media users, but we have a responsibility to apply properly devised methodologies to surface these,” Sambuli concluded.
For more blogs form CPI Financial, Visit our website.http://www.cpifinancial.net/blog
No comments:
Post a Comment