Posted tagged ‘social media analysis’

LATI: A better way to measure influence on Twitter?

23 June 2011

Twitter hero

I’ve never been comfortable with attributing digital influence to the number of followers someone has on Twitter.

To me, it’s more a measure of your longevity on the platform. The longer you have been on Twitter, the more followers you will have collected over the years.

Sure, the quality of your tweets and other variables will have an effect, but simply comparing the raw number of followers among tweeps is not really comparing apples with apples.

Three fresh green apples and one orange

I was ruminating over this when it dawned on me: why not divide the number of followers by the number of years the person has been on the platform? That will remove the variance due to longevity from the equation.

For example, I currently have 831 followers and my Twitter age is 2.1 years, so my Longevity Adjusted Twitter Influence (LATI) is:

831 / 2.1 = 396

According to convention wisdom, someone who has 1500 followers is much more influential than I am. In absolute terms that may be true, but if their Twitter age is 4 years, their LATI is 375 – which suggests I am relatively more influential than they are. That means I’m on track to becoming more influential overall.

Compare that to someone who joins Twitter and attracts 200 followers in 3 months. That’s a LATI of 800 which blows both of us out of the water.

Clock

In the short term I imagine a typical person’s LATI would follow an s‑curve, whereby they take a while to attract followers in the beginning, then they ramp up as word spreads, then they plateau out again as their target demographic is exhausted. Over time, their LATI will decline as the years rack up without significantly more followers.

In contrast, truly influential people will continue to attract followers into infinity, so their LATI will remain high.

Einstein sticking out his tongue

Now I’m no mathematician, so my logic may be all screwed up. But to me it’s more meaningful because it levels out the playing field.

Of course, the metric doesn’t recognise who is following you. Someone with 10,000 followers won’t be very influential if those people have neither the means nor the inclination to act upon their pearls of wisdom.

Conversely, someone with only 3 followers will be incredibly influential if those people happen to be the President of the United States, Rupert Murdoch and the Head of the European Central Bank.

So notwithstanding complicated and opaque measures like Klout, LATI provides an open and convenient snapshot of digital influence. At the very least it’s fun to toy around with.

I stand corrected

8 November 2010

I wasn’t impressed at this month’s SMCSYD meeting.

I mean, it was professionally organised and delivered as always, but I just didn’t buy into what the presenters were saying.

The topic was How can social media analysis help predict results? and several highly regarded social media strategists recounted their work in mining Twitter and other online forums leading up to the Australian election.

The central message was: Look, there are heaps of voters on Twitter. If we can sample their sentiment, and combine that data with how well we think the two major parties are campaigning online, we should be able to predict who will win the election.

Twitter birdI didn’t just scoff, I tweeted:

I don't know about this. Most twitterati are staunchly leftwing. Will twitter use really affect their votes?

My rationale was that if most people on Twitter lean heavily to the left, so will their vote, and nothing stated or debated on Twitter will change that. Therefore the Twittersphere is not representative of the broader citizenry, so any analysis of it will be redundant.

Someone following the #smcsyd hashtag politely challenged my assumption that most Twitterati are staunchly leftwing. Fair enough, but when a photo of Tony Abbot in his budgie smugglers elicited sneers and giggles from the audience, I knew I was on the right track.

Then something unexpected happened… one of the presenters showed us how his analysis predicted an even result. This is impressive not only because I didn’t see it coming, but also because the election result was so tight that it produced a hung Parliament.

Whooshka!

How did that happen?!

Was Tony Abbot’s online engagement so effective that it shored up his stats against the lefty chatter? No – the presenters claimed his social media campaign was poor.

In that case, if my assumption was correct, the analysis should have predicted a landslide victory for Labor. The fact it didn’t happen could only mean one thing: my assumption was wrong.

Ass

OK, I decided to take a dose of my own medicine and collect some facts. So I ran a strawpoll comprising one simple question:

How left-wing or right-wing are you?

I provided five response options on a Likert scale: Very Left, Left, Center, Right or Very Right.

With some retweeting help from my twiends, I received a grand total of 20 responses. Not very scientific, I know, but here are the results nonetheless:

Strawpoll results: Very Left 10%, Left 35%, Center 15%, Right 40%, Very Right 0%

As you can see, my tiny sample of the Twitterati is uncannily balanced. Not only does a substantial proportion of the population consider itself politically centered, but a large proportion considers itself right-wing.

So why did I think the Twitterati were a bunch of tree-hugging GMF-fearing border-opening closet communists?

Upon reflection, I think one reason is that the only political tweets I ever seem to see are left leaning. Why that is the case, I do not know. It might just be coincidence.

Another reason, however, is that the Q&A TV show seems to televise mostly left-wing tweets. Heaven forbid I accuse Australia’s ABC of bias, although others more illustrious than me have done so in the past. More likely Q&A attracts a strong left-wing following, and I suppose that subconsciously influenced my view of the Twitterati in general.

Cartoon straightening out the left bias of the ABC's Bananas in Pyjamas

Anyway, something that one of the SMCSYD presenters said that I whole-heartedly agree with is that the Twittersphere is a niche demographic.

This view is supported by the analysis which isolated the hottest subjects of discussion as being the National Broadband Network and the proposed Internet Filter. Of course people who spend time online are going to have a heightened interest in these issues.

So while the rest of the public probably couldn’t care less, Twitter is clearly an important battleground in the war that is politics.

Pollies take note!