Comments on: What gets you Twitter followers? Part 1: profile usage

By: andrewchapman.info – What gets you Twitter followers? Part 3 of 3: content

andrewchapman.info – What gets you Twitter followers? Part 3 of 3: content — Wed, 23 Dec 2009 11:27:33 +0000

[…] Part one looked at user profiles. Generally, the more you fill out your profile (description, avatar, background image etc), there seems to be a correlation with increased number of followers; and high-status description terms (’entrepreneur’, ‘author’, ’speaker’ etc) perform better than, er, low status ones (’student’, ‘nerd’ etc). […]

By: andrewchapman.info – What gets you Twitter followers? Part 2: friends and frequencies

Thu, 17 Dec 2009 11:24:36 +0000

[…] tool (tracks your Twitter stats over time, and analyses your tweeting style and personality). In Part 1, I looked at how people’s profiles might correlate with their number of followers, and a few […]

By: hatmandu

hatmandu — Wed, 09 Dec 2009 08:38:16 +0000

Rick: interesting – I’ll see if my data corroborates that for part two.

By: OK! OK! I’ll blog « Writer Way

OK! OK! I’ll blog « Writer Way — Tue, 08 Dec 2009 22:34:05 +0000

[…] Part I of “What Gets You Twitter Followers,” Andrew Chapman of Hatmandu.net analyzes the profiles of thousands of […]

By: Rick

Rick — Tue, 08 Dec 2009 21:11:47 +0000

From my observations it appears that there is a strong correlation between the number of twwets and the number of followers. The ratio seems to be in the ballpark of 1:1

By: hatmandu

hatmandu — Tue, 08 Dec 2009 15:47:32 +0000

Yes, it has occurred to me that I should probably have filtered out the top end (there are half a dozen accounts with more than a million followers; 107 are over 100,000 (including those over 1m); 684 over 10,000 (etc); 4947 over 10,000; n = 49981). The median is only 92. But I wasn’t really sure what would be a ‘fair’ cut-off point for my dataset. This probably needs a better statistician than I am! But the comparisons might be useful, even if the ‘average’ figure is a bit misleading. Advice welcome!

By: robinhouston

robinhouston — Tue, 08 Dec 2009 15:28:03 +0000

Your average follower counts are surprisingly high. I wonder how they’re distributed. It would be interesting to see a graph of the distribution. More simply, is the median very different from the mean? I wonder if the mean is skewed by a small number of accounts that have an unusually large number of followers.

By: hatmandu

hatmandu — Tue, 08 Dec 2009 13:13:09 +0000

Actually, no, this is a different project! (The corpus things is here.) Anyway, interesting comment - and thanks for pointing out that the word everyone needs to remember here is 'correlation'!

By: J-P

J-P — Tue, 08 Dec 2009 12:45:31 +0000

So that’s what you’ve been up to with your corpus.

I’d love to see this data inspire intervention studies. For example: take 1000 accounts with varying numbers of followers, and add/remove homepage URLs randomly from them, and see if there’s a statistically significant correlation over time in the change of their follower numbers.

Amazon do this A/B testing all the time; maybe Twitter could quietly drop/reveal URLs on different profile pages at random to test this. Maybe they already are doing.