Trends on #ausvotes during the Australian Election, Pt. 2

So, on to part two of our examination of trends and patterns on the #ausvotes Twitter hashtag during the 2010 Australian federal election campaign. (Part 1 is here.)

In the following posts, I’ll be interested to chart the rise and fall of specific themes during the five weeks of campaigning that we’re examining here, and to do so I’ll largely follow the approach I’ve used in Part 1 for charting the volume of mentions of the two leaders in #ausvotes tweets. But to get there, we need to work out what were key themes during the campaign, at least as far as coverage on Twitter was concerned. To get a clearer picture of that, I’ve run the more than 400,000 #ausvotes tweets we’ve captured through Twapperkeeper through the content analysis software WordStat, which provides an overview of both individual keywords and multi-word key phrases found in the data. Here are the top 50 results for each:

  FREQUENCY     FREQUENCY
ABBOTT 37578   TONY ABBOTT 11707
ELECTION 33488   JULIA GILLARD 6182
VOTE 30982   HUNG PARLIAMENT 4969
LABOR 28114   WYATT ROY 4231
GREEN 24353   MOVE FORWARD 3277
GILLARD 23942   ABBOTT IS 3138
TONY 22949   STOP THE 3125
JULIA 18026   FAMILY FIRST 3051
ALP 15089   CLIMATE CHANGE 2654
PEOPLE 14221   TALK ABOUT 2652
LIBERAL 11896   MARK LATHAM 2651
CAMPAIGN 11764   SEX PARTY 2609
TIME 11656   PRIME MINISTER 2588
POLICY 11548   BOB BROWN 2487
PM 10807   BOB KATTER 2425
NBN 10442   KEVIN RUDD 2405
GOOD 10349   WATCH SATELLITE FROM YOUR PC 2321
QANDA 9568   GAY MARRIAGE 2004
WIN 9485   AUSTRALIAN ELECTION 1831
COALITION 9415   STOP THE BOAT 1739
PARLIAMENT 8857   LIBERAL PARTY 1654
HUNG 8702   MR ABBOTT 1585
RUDD 8277   SEAT OF 1539
ABC 8207   ELECTION CAMPAIGN 1538
JULIAGILLARD 7594   INTERNET FILTER 1502
JG 7436   BOAT PEOPLE 1482
LIBS 7423   STAND UP 1424
MOVE 7098   GILLARD IS 1354
BOAT 7012   MAJOR PARTY 1300
GOVERNMENT 6988   FEDERAL ELECTION 1291
LATHAM 6978   MAXINE MCKEW 1262
REAL 6755   ASYLUM SEEKER 1194
STOP 6685   ELECTION DAY 1193
FAMILY 6605   MR RABBIT 1179
VOTING 6566   LABOR PARTY 1175
WORK 6288   VOTE LIBERAL 1167
POLL 6191   SAUSAGE SIZZLE 1141
FORWARD 5967   MOVE AUSTRALIA FORWARD 1130
SUPPORT 5762   REAL ACTION 1123
ROOTY 5342   POLLING BOOTH 1099
LATIKAMBOURKE 5317   ABBOTT SAY 1081
TWITTER 5255   ANTONY GREEN 1051
LEADER 5253   CHANNEL 9 1046
LOVE 5252   ELECTION COVERAGE 1022
WYATT 5230   TONY ABBOTT IS 1015
VOTER 5212   INCEPTION DVD QUALITY 999
BIG 5208   REAL JULIA 998
CHANGE 5194   FORWARD TO 993
FILTER 5132   SKY NEW 991
ROOTYQ 5084   WATCH SATELLITE FROM YOUR PC WYATT ROY 990

 
Note: I’ve removed various meaningless or overly generic terms from these lists. For the keywords, this includes ‘bit’ and ‘ly’ (from URL shortener bit.ly), ‘party’, ‘Australia(n)’ (which both have too many possible meanings to be useful), ‘seat’, ‘make’, ‘day’, ‘today’, ‘won’, ‘question’, etc.; for the key phrases, this includes ‘vote/voted/voting for’, ‘has/have/had been’, ‘time to’, ‘lot of’, ‘election is’, ‘watch the’, ‘Australia has/had/have’, ‘sound/sounds/sounded like’, ‘live/lives/lived in’, ‘listen to’, ‘make a’, ‘form a’, etc. While such removals are necessary to focus on the most meaningful content, they do also remove some potential meaningful data – perhaps most notably, any mentions of the newspaper The Australian, whose name contains not one but two ‘generic’ words. Unfortunately, short of engaging in extensive manual coding of the data (which, with 400,000 tweets, I’m not keen on), there’s simply no way to reliably distinguish between ‘the Australian’ meaning ‘The Australian’ and other uses in phrases such as ‘the Australian people’, ‘the Australian government’, ‘the Australian election’. Also worth noting: WordStat tends to reduce words to their basic form – so ‘Abbott say’ is most likely ‘Abbott says’, ‘Sky new’ is actually ‘Sky News’, etc.

Some immediate observations from these lists: as we’ve seen in Part 1, mentions of Abbott win out over mentions of Gillard, whichever way we slice the data. Part 1 presented cumulative totals on the number of tweets which mentioned any of the words ‘Tony’, ‘Abbot’, ‘Julia’, or ‘Gillard’, but even if we narrow those criteria to mentions only of ‘Abbott’ (in the correct spelling) vs. ‘Gillard’, or ‘Tony Abbott’ (as a complete phrase) vs. ‘Julia Gillard’, Abbott comes out on top.

Against this, however, we should also note that where mentions of their respective parties are concerned, the balance is reversed. ‘Labor’ (28114) and ‘ALP’ (15089) rank more highly than ‘Liberal’ (11896), ‘Coalition’ (9415), or ‘Libs’ (7423), and the Queensland version ‘LNP’ appears only in 3108 tweets (I’m not counting ‘National’ here, since the word could both refer to the National Party or be used in any number of other, unrelated contexts). This could be a sign of the complex multi-party structure of the Coalition (combining Liberals, Nationals, LNP, Country Liberal Party, etc.), but even adding up the numbers above the Labor side wins 43203:31842 over the Coalition, so there is a sustained trend here. And again, as with the leaders, the volume of tweets does not imply agreement or disagreement with their political positions, of course – it merely indicates the Twitterati’s overall level of attention. And while ‘Green’ also appears prominently in the list of keywords, a little caution is indicated: in addition to referring to the Australian Greens, it could also have been used in mentions of the ABC’s election analyst Antony Green!

At any rate, in combination these lists of the most tweeted terms and phrases point us to a number of worthwhile trends to chart over the course of the entire campaign period. Some of them we can already assume to be late entrants: the idea of a hung parliament, or the focus on Wyatt Roy (the successful 20-year-old Liberal candidate) and on Bob Katter (the independent MP who now finds himself amongst a handful of independents holding the balance of power) are more than likely to be phenomena to emerge on election night and in its immediate aftermath; Katter and Roy in particular were not major themes during the preceding capaign. Others we would assume to have been more persistent throughout the five weeks – discussion of the two parties’ national broadband plans, of Labor’s Internet filter agenda, of policies on asylum seekers or climate change were all prominent at various stages of the electioneering process. So, over the next couple of posts, I’ll chart these in some more detail.