Did Twitter manipulate the fake account numbers? I think so
In the latest tweet, Elon Musk of $Tesla (TSLA.US)$ mentioned that $Twitter (Delisted) (TWTR.US)$ uses a sample size of 100 to determine if the proportion of accounts that are fake.
![Did Twitter manipulate the fake account numbers? I think so](https://ussnsimg.moomoo.com/1652722865503-random6515-101592816-android-org.jpg/bigmoo)
Twitter have tons of smart people working for them, but to choose a sample size of 100 out of 330 million monthly active users is ridiculous. Even someone with basic statistics knows that. So I'm guessing they are cooking the books, and blatantly cherry picked a method that suits their narrative.
With a total user base of 330million active daily users, using 100 as a sample size will result in a margin of error of 13% (99% confidence interval). Given that Twitter mentioned their fake accounts are <5% of the total user base. They could be off by 13%. Which means the fake users could be as high as 18% of the total user base.
This is within the 5 to 30% range that Sparktoro found. Elon could just use this tool to find out the proportion of fake followers he have though instead of analysing it himself (https://sparktoro.com/tools/fake-followers-audit)
Using the sample size table, Twitter should have used more than 16,586 as a sample size instead if they like it to be very accurate (1% margin of error) or maybe even 2,654 for a 2.5% margin of error.
But 100 is ridiculous. That's why there is a saying that there are "Lies, damned lies, and statistics".
With a total user base of 330million active daily users, using 100 as a sample size will result in a margin of error of 13% (99% confidence interval). Given that Twitter mentioned their fake accounts are <5% of the total user base. They could be off by 13%. Which means the fake users could be as high as 18% of the total user base.
This is within the 5 to 30% range that Sparktoro found. Elon could just use this tool to find out the proportion of fake followers he have though instead of analysing it himself (https://sparktoro.com/tools/fake-followers-audit)
Using the sample size table, Twitter should have used more than 16,586 as a sample size instead if they like it to be very accurate (1% margin of error) or maybe even 2,654 for a 2.5% margin of error.
But 100 is ridiculous. That's why there is a saying that there are "Lies, damned lies, and statistics".
![Did Twitter manipulate the fake account numbers? I think so](https://ussnsimg.moomoo.com/1652722613369-random3169-101592816-android-org.jpg/bigmoo)
Disclaimer: Community is offered by Moomoo Technologies Inc. and is for educational purposes only.
Read more
Comment
Sign in to post a comment