Elon Musk said that in order to find out whether the fake/spam accounts on Twitter are below 5%, as the company has long claimed, he will randomly 100 followers were selected as a research sample. However, experts in social media, disinformation and statistical analysis say Musk’s analysis methods are deeply flawed.
To find out, my team will be taking a random sample of 100 @twitter followers. I’ll invite others to repeat the same process and see what they find,” Musk tweeted out his approach, adding,” Choose an account with a large number of followers, ignore the first 1,000 followers, and pick one every 10. I’m open to better ideas.
Facebook co-founder Dustin Moskovitz commented on the issue via his Twitter account. He pointed out that Musk’s method was not actually random, used too few samples, and there was huge room for error. “I think ‘not trusting the Twitter team to help with sample extraction’ is a red flag in itself,” he said.
University of Washington professor Carl T. Bergstrom has co-authored a book to help people understand data and avoid being deceived by false claims online. He said sampling a sample of 100 followers on any one Twitter account should not be considered “due diligence” for a $44 billion acquisition.
The 100-person sample size is orders of magnitude smaller than the norm for social media researchers studying this sort of thing, he said. The biggest problem Musk faces with this approach will be selection bias. “There is no reason to believe that followers of an official Twitter account are a representative sample of accounts on the platform.
Maybe bots are less likely to follow this account to avoid detection, maybe they are more likely to follow so as to appear to be normal accounts. Who knows? But I just can’t understand what Musk has done other than deceive us with this stupid sampling scheme.”