What twitter followers tell me about my election candidates
A quick experiment before I vote …
It is provincial election day here in Ontario, Canada. My thoughts are a little divided about how to vote. The party I normally vote for isn’t impressing me, and the incumbent party is one that I could never vote for — mostly because of my background.
So I was looking for ways to help me decide, and I thought that I’d turn to Twitter to find out how the candidates are using it, who has the best presence, what their following tells me.
First of all I searched for their accounts. It was tricky to find some of them, but I eventually found a tweet where they wished each other luck …
You can see from this overview that there are around 100,000 voters in my riding — yet the follower count for the candidates is very low. It surprised me.
So that can mean some of these things:
- Twitter users in my riding are not that into politics.
- Twitter is not a priority for these candidates.
- Twitter is not a priority for voters in my riding.
I would have expected Twitter to form part of the communication of any of these candidates platforms.
Follower lists fascinate me. I have a theory that followers follow out of respect for a Tweeter. They want to hear more of what that person has to say. In my theory the followers are a reflection. We are what we Tweet — so I have been devising a tool to help me learn more about the followers of a person.
My tool samples up to 1000 followers, and forms a word cloud from the bios of a Twitter user. The bigger words in the cloud are the words that occur most frequently in the followers biographies. The smaller words are the ones that occur less frequently and the scale runs between.
The data pool is 132. We can see that Ottawa, ndp, and candidate form the biggest words — so this makes me think that a lot of the city ndp candidates follow John.
Other words I see — Carleton, univeristy, school and health. So perhaps some university student followers.
The data pool is 1000. We can see that Ontario, Ottawa, conservative, and candidate form the biggest words — so this makes me think that a lot of the conservative candidates follow Jack.
Other words I see — Politics, business, association and advocate. Curiously the word ‘junkie’ reoccurs in his follower list — I tweaked my software to confirm — this word was preceded by ‘political’ or ‘cdnpoli’ — but it made me smile about the method behind my cloud.
In any case — a lot of political followers.
The data pool is 93. We can see the same pattern for her party name.
What impressed me more — despite the small data pool — is that Rosalyn has a lot more ‘humanity’ in her cloud than the others. music, world, football, good and crucially, the name of my town — Kanata and neighbouring town Stittsville.
The data pool is 29. We can see the same pattern for his party name.
There is more interest too, in real world items — health, healthy, homelessness, food, runner.
Who would you vote for based on this? I mean, this is not my recommended way for choosing a candidate — but it is a way.
I can tell you that it did influence me in my vote.
Follow up … @Kathleen_Wynne won the election as the new Premier. The bios of her followers paints a refreshingly politics light picture — with words like ‘life’, ‘love’ ‘enthusiast’, ‘social’, ‘aspiring’ and ‘human’.
You can try this tool yourself — runs on IBM Bluemix using Node.js, and stored in IBM DevOps Services. I’ll write more about it soon.
