#1

Hi everyone,

I’ve looked around a bit on the forums here and on the internet but I can’t seem to find the answer to my specific question. I’m sampling all tweets that contain a word from a list of hate speech keywords I’ve assembled, and then querying all geotagged tweets containing those keywords within specific parameters. I’m seeing something quite peculiar happening. I know there’s supposed to be a massive drop in location usage in April 2015 due to changes in the way location was recorded (I’m fine using place_id here instead of specific lat-long coordinates), but it’s instead showing up in September 2014:

image
image805×931 43.6 KB

Is there any insight as to what’s happening here? I don’t see anything in historical news searches or in Twitter’s developer documentation that has clarified what’s going on. I’ve also double-checked the timestamps on the raw data from my pull–these are all from the right dates, and UTC couldn’t possibly do any month-level shifting.

Thanks in advance.

1 Like
#2

I’m not able to comment on this specific change, but I can comment on Tweet geo overall:

  • per a conference around (but very vaguely) that time, it was stated that only 1-2% of Tweets carry geo information
  • certainly from around the same time, perhaps more in the 2015-8 timeframe, the Twitter app switched from specific geo data to places, and to defaulting geo to off.
1 Like
#3

This is super useful, thanks! Any chance you can pinpoint the exact day this happened?

I’ve a bunch of other changes but don’t have this one:

1 Like