So you want to do a twitter location search. You’ve found the complete guide.
The combination of social media and mobile technology has resulted in an increase in available data about what people are thinking and doing in a specific location.
Here are some resources I’ve developed to help people who wish to monitor or search Twitter by location. These resources cover methods that are more granular and controllable than Twitter’s “near” search operator.
Twitter location search resources:
- How to find a latitude/longitude
- Use Hootsuite to search Twitter by location
- Use Tweetdeck to search Twitter by location
- Example Twitter location search strings
Limitations to social/local data
As exciting and useful as this location social data is, there are a number of limitations which should be kept in mind by anyone who wants to know more about the social chatter at a specific place.
Twitter location search limitations: technology
The method outlined in the Twitter location search tutorials listed above is the most direct way to query Twitter for geocoded data. It is not perfect, however. A Tweet is tied to a specific location based on the following:
- Twitter user has location enabled for Tweets
- Twitter user has a “from” field in their profile that has a specific town
These items are checked sequentially, near as I can tell. So if someone has location enabled from their cell phone and uses Twitter that person’s tweets will be tied to specific GPS coordinates.
If someone doesn’t have location enabled on their cell phone the next thing checked is the user’s profile. If there is something entered into the profile then Twitter will attempt to tie all of that person’s tweets to the location entered into the profile.
This system is good and allows users a great degree of latitude in terms of their privacy. People using location data from Twitter, however, should be aware of possible gaps and noise in the data that results from this setting (which again, I personally find strikes a useful balance between utility and privacy):
- A large audience will have neither a location nor location enabled cell phones. These Tweets will not show up via any publicly accessible twitter location search method. This represents a gap in the data.
- The matching of a location in a user’s profile to a location in the real world is often not entirely accurate. For example, if someone in Manchester UK fills out their profile info with “Manchester” their tweets can show up in a geocode Twitter search around the coordinates of Manchester NH USA. This can introduce noise into the data.
- Also, when someone has location disabled on their phone but does have a home location in their Twitter profile inaccuracies can be introduced to the data if that person leaves on vacation. For example, if my location is Burlington VT and I have location services disabled and then go to Whapeton ND my tweets will still be tied to Burlington VT.
I don’t believe that these gaps, noises and inaccuracies completely degrade the value of the data. But anyone using or relying on Twitter location search data should be aware of these issues and have methods for attenuating their data and insights accordingly.
Twitter location search limitations: human factors
Another limitation to the data you can get by searching Twitter by location is related to people–what geeks like me call “human factors.” Since the data we are examining in these location searches is inherently social, how people communicate will have a large impact on the available data.
The issues here will involve exclusion and biasing data:
- People who don’t have access to digital technology or have limited access to digital technology will not be represented in the data gathered for Twitter location searches. The result is an exclusion in the data of this portion of the population.
- Similarly, some people who do have access to the required digital technology will choose not to participate in social networks like Twitter. Or they may participate in Twitter but choose to disable location on their mobile devices and also leave their location empty in their Twitter profile. Data for this group will not be present either; this is an exclusion.
- Of the people who have access to technology and choose to give location signals to Twitter some will be more chatty than others. Some people will simply use Twitter to read other people’s streams. Some people will blab on and on and on all day. There will be more data available for the people who are chatty than for the people who don’t post very often. This will bias the data set towards the digital extroverts.
As with the technology limitations I don’t think the limitations associated with human factors invalidates or destroys all of the value of location-based Twitter search monitoring. But anyone relying on this social/local data set should be aware of the human factor limitations and attenuate accordingly.
Why discuss Twitter location search limitations at all?
The tendency of those who believe in the usefulness of technology is to be a constant cheerleader. They’ve learned this tactic because there exists an opposition group: the naysayers.
Between the cheerleaders and the naysayers there can be incredibly valuable information. But if we leave it to the cheerleaders the chance exists that we will overestimate or simply believe in technology press releases. And if we leave it to the naysayers we will not try anything new or make new discoveries.
If we can have a clear, honest assessment of a technology and the limitations of its use then we can develop more meaningful insights. This sort of discussion requires us to acknowledge various shadings of value, bias, urgency and so on.
Especially when dealing with data and information as it relates to humans, this kind of data is rarely an on/off switch. It isn’t black or white. It isn’t positive sentiment vs negative sentiment. There are shades.
If we are aware of the shading and can discuss them openly within our specific organizations with regard to our specific needs, then location search data for Twitter can become a valuable tool for understanding human behavior as it relates to technology and location.