3 ms·

He did make his case - 395 racist tweets out of 120 million voters is practically background noise. You should expect all manner of extreme behavior on a sampl

He did make his case - 395 racist tweets out of 120 million voters is practically background noise. You should expect all manner of extreme behavior on a sample size that large.

Anechoic 14y ago
Again, that 395 number is not the total number of tweets, it's the number of tweets that included geolocation data. That 395 was likely a subset of the total number of racist tweets.
- cobrausn 14y ago
  And if it were 1,000? 5,000? 25,000? You're still talking epsilon-level percentages here.
  - jdunck 14y ago
    In some previous twitter work, I've seen geo-located tweets at about 2-3% of all tweets (assuming no bias), so charitably 19750 racist tweets here. Which is to say, I agree with you.
  - mistercow 14y ago
    "Epsilon-level percentages" isn't really a useful concept. As a sample size grows, phenomena become significant at smaller percentages. The set of all tweets is gigantic, so we can still make significant observations based on a "small" subset. The point of this article is not to say "look at how many racists there are on Twitter." The point is to say "Let's look at the racist people on Twitter. What is the geographical distribution?" 395 geotagged tweets is absolutely enough to start making some interesting observations. Those observations can only be about relatively broad regions at that sample size, but they are interesting nonetheless.
glesica 14y ago
The sample was from a subset (geolocated tweets) and involved only a very specific set of keywords. In other words, this was not an exercise to see how many racist tweets there are. It was an exercise to see where the racist tweets generally come from. All that was needed here was a relatively small random sample. So the analysis, at least in terms of the sample size itself, is almost certainly valid. You may have misunderstood the point of the post...

He did make his case - 395 racist tweets out of 120 million voters is practically background noise. You should expect all manner of extreme behavior on a sample size that large.

Anechoic 14y ago

Again, that 395 number is not the total number of tweets, it's the number of tweets that included geolocation data. That 395 was likely a subset of the total number of racist tweets.

cobrausn 14y ago
And if it were 1,000? 5,000? 25,000? You're still talking epsilon-level percentages here.
- jdunck 14y ago
  In some previous twitter work, I've seen geo-located tweets at about 2-3% of all tweets (assuming no bias), so charitably 19750 racist tweets here. Which is to say, I agree with you.
- mistercow 14y ago
  "Epsilon-level percentages" isn't really a useful concept. As a sample size grows, phenomena become significant at smaller percentages. The set of all tweets is gigantic, so we can still make significant observations based on a "small" subset. The point of this article is not to say "look at how many racists there are on Twitter." The point is to say "Let's look at the racist people on Twitter. What is the geographical distribution?" 395 geotagged tweets is absolutely enough to start making some interesting observations. Those observations can only be about relatively broad regions at that sample size, but they are interesting nonetheless.

glesica 14y ago

The sample was from a subset (geolocated tweets) and involved only a very specific set of keywords. In other words, this was not an exercise to see how many racist tweets there are. It was an exercise to see where the racist tweets generally come from. All that was needed here was a relatively small random sample. So the analysis, at least in terms of the sample size itself, is almost certainly valid. You may have misunderstood the point of the post...