Australian elections: The geography of Twitter
I wanted to create a map of where people who were following Australian political candidates and/or tweeting about the Australian elections were living. I’ve been getting similar data for Australian sport teams and this feels like a natural extension. I’m not doing an analysis of this data and these maps because while I feel I know enough about Australian sport to draw conclusions, I don’t feel comfortable with Australian politics. If some one else wants to do that or use my raw data, please feel free.
Just as a caveat, this is a limited data set that was collected on the morning of July 19, 2010. I used a tool to mine this and the amount of data included was based on what that tool gave me.
Data Collection Methodology
- Identify the major candidates in the Australian elections on Twitter (Tony Abbot, Bob Brown, Julia Gillard) and some of the more popular search phrases and hash tags used by Twitter users (#alp vote, #aus2010, #ausvotes, #laborfail, #myliberal, #nocleanfeed , #qanda, Abbott-Gillard, asylum boat, Gillard, Greens Canberra, Krudd Labor, Labor Liberals, Rudd Liberals, Tony Abbott, Truss LNP, Warren Truss, WorkChoices)
- Go to http://www.searchtastic.com/ , run a search for those phrases and accounts. Export results to an Excel spread sheet. (Total tweets picked up: 27,506)
- Create a new sheet. Copy and paste only those Tweets by people that list a location. (Total tweets included: 26,198)
- Using location information from profiles, try to determine the city, state, country that the person lives in. Example: Melbourne = Melbourne, Victoria, Australia. ChiTown = Chicago, Illinois, United States. Brizzy = Brisbane, Queensland, Australia. 32.38504,-83.647406 = Perry, Georgia, United States. Oz = Australia.
Methodology: Gillard, Abbot, Brown, Truss
- On Excel, turn AutoFilter on.
- Filter the Tweet column to contain one of the following terms: Gillard, Abbot, Brown, Truss
- Copy and paste the City, State, Country columns to a new worksheet. Select those columns on the new worksheet.
- Sort data by city. Remove all locations that do not include City information.
- Go to Advance Filter. Copy the Unique Records to Another Location.
- Using a COUNTIF formula, count the number of tweets that mentioned the term by city.
- Go to http://www.batchgeo.com/ and copy paste the city, state, column, count in the box for step 1. Click on the Map Now button.
- When Geocoder is finished running, select the Show Geocode button, and copy and paste that data to an Excel worksheet. Save each politician’s tweet references on its own CSV file.
- Go to http://finder.geocommons.com/ and import each csv file. Finish processing each file using latitude and longitude coordinates.
Go to http://maker.geocommons.com/ and create one map for all four politicians. Use visual theme, sizes.
Please note this map is based on total tweets by location. It is not total individuals who tweeted with the selected keyword.
Methodology: Labor, Liberals, Greens, National Party
This methodology is a repeat of the previous except the names of the politicians are replaced with the names of the parties.
Methodology: Education, Work Choices, Asylum
This methodology is a repeat of the previous except the keywords are: Education OR Schools, Work Choices OR WorkChoices, Immigrant OR Asylum, NTB or Broadband, Environment OR Global Warming, Health OR Doctor, Afghanistan OR Diggers.
I’m planning on adding two maps to this post tomorrow. One shows the location of followers for Tony Abbott, Julia Gillard and Bob Brown. The other map will include location of tweets relating to certain issues.