Full Duplex has partnered with Abacus Data to provide complementary analysis of the Ontario election. Abacus Data is conducting online surveys using more traditional polling methodologies. Full Duplex is conducting analysis of Ontario election activity using Sysomos Heartbeat and MAP, and our own proprietary analysis tool, Compass.
With the election taking place this Thursday, May 2 through June 8 will be our last period of comprehensive analysis prior to the vote. We will conduct further analysis after the election.
Today’s analysis considers some high level numbers related to content generation and participation. Tomorrow we’ll look at how the leaders and parties are faring within the chatter, including sentiment analysis and trends. Wednesday, we’ll look at issues and how they’re playing out online. You can follow our election analysis on our blog.
This week’s analysis looks at May 2 through June 8, inclusive (with a few looks at the last six days).
Twitter dominates online election (and debate) activity
Last week’s debate drove a significant spike in online activity related to the election. The nearly 42K debate related tweets combined with the rest of the election Twitter chatter on June 3 to create a spike of 57K mentions. That made debate day the most active for online chatter in the election. So far. This Thursday, Election Day, will almost likely put the debate chatter to shame.
Between June 2 and June 8 (inclusive), the Ontario election (and related people and events) was mentioned in 169,478 tweets (93.5% of evaluated online chatter), 6,024 publicly-visible Facebook posts (3.3%), 4,495 news articles (2.4%), 551 forum threads (0.3%) and 524 blog posts (0.2%). The 82 relevant YouTube videos were too few in numbers to rank. In all, there was an average of 25,879 online mentions of the election each day, up from 16,490 during our last evaluation period.
Since the election was called, there has been 604,370 relevant Twitter mentions (92.9% of evaluated online chatter), 20,899 publicly-visible Facebook posts (2.8%), 19,986 news articles (3.0%), 2,583 blog posts (0.3%) and 1,668 forum threads (0.2%). The 357 YouTube videos account for less than 0.1% of the mentions. This works out to be an average of 17,564 mentions per day between May 2 and June 1, inclusive. That’s up from 15,058 average daily online mentions noted during our last evaluation period.
As noted in the graph, below, last week’s debate day set a new peak for daily online mentions of the election. There were 57,418 relevant tweets and 1,468 publicly visible relevant Facebook status updates that day, alone.
Many (more) participants, even more tweets
Last week’s debate drove a significant increase in online participation in election chatter. In fact, all participation levels, from those who issued just one tweet and left to those who have issue eight or more tweets, saw a significant bump in participation rates on June 3. And, the growth trajectory in the lower-engagement categories remains surprisingly steep.
In all, 55,725 Canadians have issued election related tweets (that’s 11,833 new participants since last week’s analysis of 43,892 participants). With the tweet count for the election-to-date at 604,370, that works out to an average of 11 tweets per participant. Of course, that’s not the way the participation rates work out.
27,441 participants (49%) in election-related tweets have come and gone in a single tweet. That means, half of participants are not engaged in the chatter. They typically retweet comments, content or headlines shared by others and leave. These people offer analysts (and campaign teams) the opportunity to register another potentially new expression of political affinity based on direct or inferred of sentiment/stance.
There is a greater chance of determining affinity among those who tweet more often since by issuing more than one tweet, there’s the potential to discover more information.
Participation rates in each of the four analysed categories, along with growth rates from last week, are as follows:
- 1 tweet: 27,441 participants (49%), up 26% from last week
- 2-4 tweets: 14,604 participants (26%), up 28% from last week
- 5-7 tweets: 3,985 participants (7%), up 25% from last week
- 8+ tweets: 9,695 participants (18%), up 29% from last week
Those who tweet 8 or more times include those who work in the media, analysts, political players (including candidates and members of campaign teams) and political enthusiasts.
As long as there are new participants, we can expect to see new expressions of opinion and political affinity. When the the participation curve levels-off, we’ll be looking at an echo chamber where the most notable information could be declarations or change in voting intention.
This next graph features three lines. They show the number of tweets issued each day (Tweets), the number of unique contributors each day (Unique tweeters), and the trend line indicating cumulative participation over time. Notice both the spike in tweets and unique tweeters, as well as the significant bump in cumulative participation on June 3.
Like most political chatter online, the Ontario election is seeing mostly amplification in the form of retweets. This means people are mostly re-broadcasting headlines, content and “zingers” issued by other people as regular tweets containing original content of some form or another. And true to online political form, conversation represents a very small portion of the activity.
Rather than look at overall participation styles, we chose to look at the participation style at the four levels of participation. While there are some obvious consistencies among the four groups of election tweeters (such as the leaning toward amplification rather than communication or conversation), there are also some noticeable differences.
One-time tweeters are much more likely to retweet than any other category. Surprisingly, they, along with the other “less-engaged” group of 2-4 tweet issuers, are more likely to issue original content. While the most engaged participants (those issuing 8+ tweets) are much more likely to engage in conversations or exchanges with others, they are not nearly as actively engaged in conversation as one might expect.
Mostly low and medium “authority” rankings
Sysomos assigns authority rankings to participating Twitter accounts based on the size of their Twitter communities, how much content the account issues, how often the user is retweeted and how often the account engages in Twitter exchanges. Twitter authority is not always consistent with real credibility. Someone who’s particularly powerful in the political world could be completely irrelevant in these rankings based on their own Twitter patterns.
There was a small shift in authority distribution last week, largely due to the increased participation rates during the debate. During the past week, most participants in the Ontario election discussion have a low authority ranking (59%, up 1% from last week). There’s a healthy number of participants in the medium category (40%, down 1% from last week) and a very small and steady portion with high authority rankings (1%). The high authority group includes a healthy number of media organizations and journalists, and popular political players, analysts and celebrities.
Participation skews male
Consistent with general online political chatter, participation skews male. This suggests that men are more likely to voice their political opinions online than women. Between June 2 and June 8 (inclusive), males accounted for 68% of the chatter, females for 32%. That’s a 1% shift in favour of females over election chatter to-date.
Interestingly, the propensity to engage in online political chatter doesn’t always translate to the voting booth. In both the 2011 and 2008 federal elections, women voted in higher numbers than men. According to Elections Canada, “for the 2011 general election women participated at a higher rate (59.6%) than men (57.3%), and this was true across all age groups up to age 64, where men started participating more than women. This is the same pattern seen in the previous election in 2008.”
Since the election was called, #onpoli tops the list of most popular hashtag (as it has since the election was called). #onpoli is associated with Ontario politics. #voteon is the most popular election-specific hashtag; #onvotes is another. Others which didn’t rank, including #onelxn and #onelxn14.
The #cdnpoli hashtag applies to Canadian politics and #topoli to Toronto politics. #ondp, #olp and #pcpo represent the Ontario provincial NDP, Liberals and PC’s respectively. #stophudak is a hashtag used by critics of Tim Hudak.
Most popular tweet
Tomorrow’s analysis will include comparison of leader mentions and sentiment.