I’m going to be a guest of CJME Morning News on News Talk Radio, Regina, SK this morning. I’ve been invited to talk about my analysis of the digital side of the current Saskatchewan provincial election (see my earlier posts here and here). This post is to make sure I’m talking about the most current information, looking at online activity between October 10 and October 30 using Sysomos Heartbeat and MAP.

Twitter is the online water cooler

At 89%, Twitter’s overall share of the online chatter related to the election is decidedly dominant in this election. In fact, the ratio of current Saskatchewan election Twitter activity exceeds that of the final tally of Ontario election Twitter activity by 12% (recognizing SK’s current volume 9,076 mentions is much lower than Ontario’s 126,330). Saskatchewan’s 9,076 election tweets have been generated by 1,125 Twitter accounts suggesting an average of 8 tweets per account. The most active non-media tweeters are led by @DarrenSproat@LateNite72, @colewhogan and @SaskBoy.

Blog, YouTube and Facebook activity is much lower than I expected. While blogs and videos can be effective campaign tools, they’re more often associated with public conversation during elections. That is, they allow people to capture their own take on events and share their thoughts in a more meaningful way.

While the same holds true with Facebook, that social networking site can serve as an important online community-building tool and should be used by candidates to build constituencies of support which can be activated when needed. That means, building communities before you need them (e.g. between elections) rather than opportunistically (e.g. when a writ is dropped and you finally find yourself in campaign mode). As I wrote in an earlier post, trying to build and activate an online community when you already need the support is like trying to get passengers into an airplane that’s already in flight.

The peak on October 25 in this graph reflects increased activity related to the leaders’ debate.

Participation (rather than amplification) makes Saskatchewan unique

This is where Saskatchewan takes on a very unique look. It’s an anomaly, really. In all of my election research, Twitter traffic is made up of a respectable amount of original content (regular tweets) very little in the way of conversation (as identified by @replies) and dominated by what I call The Amplifier Effect (identified by retweets). Participants in the Twitter conversation about the Saskatchewan election are far more conversational. They participate more and repeat less suggesting there’s a greater level of participation and discourse.

I suspect this is made possible by the smaller volume of daily traffic. It’s much easier for the digitally engaged to join a conversation that boasts an average of 500 (or fewer) tweets a day rather than trying to keep up with a flood roughly 5,000/day as observed during the Ontario election and 12,000/day during the federal. I’ve typically observed conversation rates of 12-15% during election campaigns; 28% is unheard of.

Male-dominated conversation

Online political discussions quite clearly attract more men than women. It’s typical to see the splits identified in the Saskatchewan election; 74% men, 26% women. Of course, as my gender boilerplate goes, the challenge with gender breakdowns of digital conversations is results are based only on accounts which disclose gender information or from which gender can be credibly determined by the disclosed name. Not everyone does that online. So, the results are based on a statistically relevant sample. The number of disclosing accounts is indicated in the graphs. Perhaps more women participate using nondescript account names.

Brad Wall and The Saskatchewan party dominate online leader and party mentions

Wall is the most-mentioned leader in online conversation (54.8%, up from last week), leading the second most mentioned NDP leader Dwain Lingenfelter by nearly 33.7%. There are two important things to remember about this graph, and others such as the party mentions graph:

  1. Mentions do NOT necessarily reflect election outcome, only that Mr. Wall is being mentioned online more often than any other leader; and,
  2. This analysis does NOT consider sentiment. That means, we can’t determine from this graph if Wall’s 54.8% share of mentions among the leaders is more flattering than Lingenfelter’s 21.1% share. There are some automated tools with use programmatic logic to determine sentiment. My experience is these tools don’t negotiate the nuances of language (such as context and sarcasm) very well. People-powered review is the way to go. Having said that, I don’t anticipate going through 9,076 tweets anytime soon.
The peak on October 25 in these graphs reflect increased activity related to the leaders’ debate.

The Saskatchewan holds a commanding lead of party mentions (74%).

Education is the most mentioned election issue

The six-most discussed issues to date are:

  1. education (27.7%, remains at #1 and up 1.8% since last week)
  2. tax (17.5%, remains #2 since and up 1% since last week)
  3. healthcare (15,8%, did not rank in top six last week, displaced students)
  4. natural resources (14.3%, remains #4 since and up 1.2% since last week)
  5. family (16.2%, remains #5 though down 2.8% since last week)
  6. agri-food (11.3%, down from #5 and down 1.2% since last week)

(Note: Percentages reflect share of mentions among the top six issues identified in the chart, not among all issues tracked during the election.)

All of the above issues were discussed more among men than women, which is consistent with my observations about gender analysis among participants. Gender is identified in two ways. The first is through disclosure in social media profiles (something not everyone does). The other is through determining gender by analyzing the account holder’s name where gender can be credibly determined by the name. The result is a statistically relevant sample for analysis.

The level of overall participation by women resulted in a gender-specific sequence that differs slightly from the overall issue sequence.

  1. education (78 mentions)
  2. natural resources (41 mentions)
  3. healthcare (40 mentions)
  4. tax (32 mentions)
  5. family (26 mentions)
  6. agri-food (17 mentions)

However, participation rates of women were higher tell a different story

  1. healthcare (40%)
  2. education (31%)
  3. family (31%)
  4. natural resources (27%)
  5. agri-food (26%)
  6. tax (20%)

Related analysis

I’ve done a lot of analysis on the use of digital and social media by politicians and in political campaigns. Here are some posts you may also find interesting: