Friday, December 16, 2016

JT's "fun little questionnaire"

VoxPopLabs, the same folks who brought you Vote Compass, has taken something of a beating in public opinion for their electoral reform quiz

Originally introduced by Justin Trudeau as "a fun little questionnaire that gets into values rather than models", Andrew Coyne referred to it as "a dippy pop quiz or botched push-poll", soundly mocked for the obvious slant of the questions with "two thirds of the “values” questions and over half of those probing respondents’ “preferences” having nothing to do with electoral reform". 

Or as I like to call it, a 'what kind of animal are you?' quiz.

Asked about this on CBC, Mark Holland was at pains to point out that the government didn't pick the questions. 
No, but you did choose the parameter of the questions; VoxPops merely did the job you asked them to do.

From the government tender awarding the contract to VoxPopLabs in September :
"The key tasks include:
Provide a draft of questions for review by the Project Authority;"
The Project Authority here is presumably the Privy Council Office, described in the bid as the "End user entity".
"Conduct a robust scientific survey in advance of the launch of the interactive online application in order to identify key themes and develop classifications based on responses;
.... determine segments and develop narratives for each cluster based on the profiles that emerge from the survey; 
VoxPopLabs says they developed a large survey, field-tested it, and then asked a panel of 3,000 Canadians "the remaining questions and used their responses to generate a cluster analysis".
This is where those five much-mocked archetype categories come from.
Develop a robust model ...  to instantaneously classify participants to the interactive online digital consultation platform into segments uncovered in the survey
Ok, so your answers slot you into one of the five predetermined archetype categories.

My complaint with the above approach is that the archetypes were built on different questions asked than those asked in the final survey, and that of the five, there is no category for people who are both in favour of some version of proportional representation but against online voting. And did we really need five direct questions about online voting and mandatory voting but none on proportional representation?

So what will the Project Authority learn from all this?
Provide categories of customized user feedback and data visualization for review by the Project Authority;
This is, I grant you, more scientific than the five categories of Heathers and Zoes and Dougies and Steves and Eunices that conservative strategist Patrick Muttart categorized Canadians into for Stephen Harper back in 2005, but I don't see it getting us any closer to electoral reform. 

Not completing the survey will result in a repeat of the ERRE Liberal MPs' contention that no one is interested. However if you do the survey, game the questions by figuring out which ones give a glimmer of support to the idea of proportional representation and mark them "strongly agree", you possibly run the risk of being disqualified because, as the ERRE Liberals also complained, everyone who showed up to townhalls or answered the much superior ERRE survey was self-selective and already had an opinion.

Filling in this survey is quite a bit like strategic voting, a reason many of us want rid of FPtP in the first place. FairVote has provided a guide to answering it.

Fun fact : VoxPopLabs mydemocracy survey cost $326K. 
The entire budget for the now disbanded ERRE committee, their five months of witness testimony, their travels across Canada, and their online survey on electoral reform systems, was $300K.


Anonymous said...

Vote Compass in 2011 election.
"Prof. Kathy Brock said she has completed the quiz numerous times with varying strategies, and on each account earned a Liberal result. Over the course of five attempts, Prof. Brock first answered all 30 statements with “somewhat agree,” then with “somewhat disagree,” then with “strongly agree,” then with “strongly disagree,” and finally, she responded with the neutral response of “neither agree nor disagree.”

For the final portion of the quiz, where users rank leaders based on competence and trustworthiness, Prof. Brock said she consistently answered with “I don’t know.”

“Each time, I came up in the centre, so I would be labelled a Liberal.”

Lorne said...

Increasingly, I am suspicious of the legitimacy of surveys. My wife does them regularly for EKOs, and judging by many of the questions they ask, they seem to be more push polls than legitimate exercises to find out what people think about issues. Such manipulation serves no one's real interests.

astrom47 said...

I agree with these comments. I fill out surveys on line but I wonder about their validity. We do not know how valid they are, there is no scrutiny.

janfromthebruce said...

Good exposure on who actually provided the parameters of focus of the questionnaire/survey, and how the archetypes were created when used on different questions in the field.

I think a good question to ask, is how many people did not receive a postal card. That card did not arrive in my mail yet.

I wonder how many households across Canada did not receive one?

Blog Archive