I thought I'd share a recent study into Twitter usage habits, conducted by Carolin Gerlitz and Bernhard Rieder. I missed this back in May, when it was first released, so apologies if you've already seen it.
The findings are significantly different to an older study from 2010 by a Microsoft team (Boyd, Golder and Lotan). This may be due to a different - arguably more robust - sampling method, using the Twitter Streaming API. Or, it may be that usage habits have evolved in the intervening three years.
The full research is available here. It is a rather dense read, though a rewarding one. For those of you with TL;DR syndrome I have extracted some highlights.
The study captured 4,376,230 tweets, sent from 3,370,796 accounts.
Hashtags appeared in just over 13% of tweets. Only 25.8% appeared more than once, and only 0.7% (1,684) more than 50 times.
Only 11% of all tweets contained URLs, which is less than half as many as stated by the Microsoft team. Social and content platforms are unsurprisingly the most popular destinations.
YouTube ranks first, though Facebook would be number one if apps.facebook.com was included. The next three most commonly linked to sites are Instagram, Ask.fm and Twitpic.
People love to talk, it seems. 57.2% of tweets contained an @username mention, vs just 36% in the Microsoft study. Some 46.8% of tweets started with an @username.
How people access Twitter
85% of tweets are sent by 15 'dominant platforms'. Twitter.com is the most common way of posting tweets (22.6%), followed by the iPhone (21.7%), Android (16.3%), Blackberry (11.5%) and mobile web (2.2%).
LA Times is the top news site, followed by the BBC, Reuters, Washington Post and ABC News.
60% of all tweets have English language settings.
Tweets are delivered via the API about two seconds after they are sent.
Some spambots manage to post over 2,500 messages in a 24 hour period - well above Twitter's 'limit' of 1,000 tweets per day, per account.