Comments on: What gets you Twitter followers? Part 2: friends and frequencies

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 11:28:33 +0000

[…] Part two discussed friends counts, and frequency of tweeting. There is an unsurprisingly close correlation between the number of friends you have and the number of followers; and you’re better off tweeting less than 30 times a day to avoid putting off followers. (Remembering always that correlation doesn’t mean causation, fact fans!) […]

By: hatmandu

hatmandu — Thu, 17 Dec 2009 18:42:44 +0000

Oh, and Matthew: most of my friends do not follow me – dare I say it’s poor statistics to extrapolate from a single example? 🙂

By: hatmandu

hatmandu — Thu, 17 Dec 2009 18:41:51 +0000

@doccosmos (Matthew):
I can’t make any claims for that line of ‘best’ fit, given that (a) I’m out of my somewhat shallow statistical depth and (b) it was generated by the graphing software, as explained. But the correlation between friends and followers is self-evident. I pointed out in part one that such things don’t guarantee causation – though on this case I think it’s self-evident that if you follow lots of people, they’re generally more likely to follow you back. And in fact all of the ‘get followers quick’ Twitter marketing schemes and scams one sees advertised work on that premise.

@nycteris:
Thanks! Yes, not hugely illuminating. I think part one had some general discoveries of interest; let’s see what part three brings! It will hopefully show whether tweeters who post endless links, for example, put people off or not.

Though as always, it depends…

By: Matthew

Matthew — Thu, 17 Dec 2009 18:20:49 +0000

“Roughly speaking, and unsurprisingly, there’s a one-to-one relationship between friends and followers. Want followers? Make friends.”

This statement is based on a correlation coefficient (R) of 0.38515 (R^2 = 0.14834) and is an example of bad statistics. There is no (statistically) meaningful relationship here – it’s just the result of least-squares linear regression on this data set. There are probably much better fits that could be had, e.g. weighted, sigma-clipped or log-linear. Also correlation does not imply causation – most of my friends do not follow me (and wisely so).

By: Nycteris

Nycteris — Thu, 17 Dec 2009 18:03:02 +0000

Interesting, but, as you said, fairly unsurprising. Can’t wait to see the next part of your study! But I don’t know that I will change my personal tweeting/Twitter style based on anything you show me.