JanRezab / Blog

Founder & Executive Chairman of Socialbakers.com

Monthly Archives: April 2013

How should journalists interpret social data?

Published: April 29, 2013

For a few months, some journalists have been misinterpreting some of our data. Especially around Facebook statistics (note: Not around pages, those are exact statistics, but around countries). We have been openly thinking if we actually just drop that section, but it would appear scattered on many other places online. We have been always pretty clear about all of our stats, and we know whats exact (numbers on different pages), and which are estimates (Facebook country stats), or which are panel based (Fake follower check), and always been clear to provide data information and methodologies.

So if you are a journalist, and receive data, you should follow some basic rules:

  • Check Source – Data, footnote, proof, reliability of source.
  • Check Sample size – If its a panel, or a set of pages, is the sample size big enough? I have seen too many reports for example on a set of a couple hundred pages – thats not good enough. When Socialbakers provide reports on pages, they are typically (unless we say otherwise), on all pages in that selected industry, and the categorization and the list of those pages is always provided publicly on Socialbakers.com .
  • Check Methodology / Footnotes – Check the data footnotes. If it says they are estimates for advertisers not aimed at journalists, its best to read that.
  • Check stats around that market / metric – If we said 3 months ago (repeatedly), that having a 50% Facebook penetration from population (note: Not online population, actual population), is AWESOME, and there is no question about that, and that has to fluctuate up and down, because you simply won’t create more people, people don’t read that. Of course, thats not interesting, thats not scandalous.
  • Contact us – Its always better to contact us before publishing anything, thats the best way to approach it. If we dont get contacted, its hard to react to anything. For example today, we have only issued a statement and I sent a few Tweets.

If all of these things are checked, then you should move on and have a great story. But “Justin Bieber has fake followers“, if we say they are fake/empty/or inactive. You should really check if they are really fake or empty, which one it is, and most importantly, if you are not sure, then contacting us is the best option.

Specifically, for the second time, its Guardian publishing these stats and using them out of context. For the second time. I hope journalist education around data is going to be better.