Live blogging from MRIA’s #NetGain8 conference in Toronto. Any errors or stupid jokes are my own.
- counting counts
- the loss of the long form census was disappointing because that’s a loss of a lot of counting that serves a lot of different cultural and community organizations and groups
- “it’s all about the sample” – that’s what she wants on her tombstone 🙂
- 20% of Canadians live in cell only households – much research excludes these people from research
- clients want yesterday’s data today so they can react in time for this evening
- online panels are not a panacea [uh, does someone really think that?!?!?]
- opt-in non probability panels are not reflective of households, yield low and unknown response rates, subject to coverage, non response, and selection bias that can’t be corrected for, should not report margin of error [and yet we still do]
- need to do dual frame sampling of both landline and cell only
- need better randomization of response categories
- need experimentally optimized call backs with replacement
- to say you use mail says nothing about the quality and details of your processes – one mail study is completely different than every other mail study
- [frank was too passionate in his talk and I missed taking notes on some of it. sorry. you missed out. 🙂 ]
- HD-IVR gets really close to census results on demos and newspaper habits
- key reason is that all adults are recruited and surveyed on the same platform
- there are more differences between research method than between landlinehousehold and cellphoneonly
- modeling up to census numbers doesn’ t work, you need better methods [completely agree! get your sample right from the beginning and you won’t have to “fix” things afterwards]
Big data – Can the use of ‘big data’ eliminate the need for yet another traditional Census in 2021?
Keith Dugmore, Director, Demographic Decisions
- About a third of people in the room answered the last census online [I’ve only answered on paper]
- 94% completion rate even though it’s compulsary to answer the census
- All the data is free [are you listening Canada and US, free!]
- Census hasn’t really changed since 1961 – household forms on paper, some postal innovations in 2001, online options in 2011
- Increasing costs and difficulties of traditional census means they need to consider alternatives
- full census to everyone
- Rolling census over 5 or 10 year period
- Short form to everyone and short form to some
- and others
- Data sources – patient data, election data, school data, DVLA, maybe even customer databases, loyalty card data
- Each one might miss miss migrant worker dependants, international students, asylum seekers, expatriats, deceased, newborn babies, some duplicates, home schooled children. But across everything, nearly everyone should be covered.
- Problems – updating may not be great, not representative samples, biases by region
- Electricity has 100% coverage, gas 80%
- Telecoms – only have address for 50%
- Sharing of this data may be prevented due to reputational risk or the data protection act
- The quality of the census is great for a day and then gets worse and worse after that. An alternative method might have greater, longer lasting accuracy. It could be done annually or quarterly. We might even get new data like income. [um… now you’re starting to push it. that’s getting really big brother.]
- We want an alternative method but is there one out there? A recommendation will be made by September 2014. [I think a recommendation on that data will be out of date in September 2013]
- The 6 Worst Market Research Mistakes #MRX (lovestats.wordpress.com)
- A Halloween chocolate lesson in nominal, ordinal, interval, and ratio data #MRX (lovestats.wordpress.com)
Robert started the session with his favourite 2010 US census story. One canvasser was assigned to a particularly rural route. She followed the driving directions and ended up going down one tiny dirty road after another even tinier dirt road. At one point, the road just stopped and turned into the tiniest walking trail. So, she picked up her briefcase and started walking. Eventually, way in the distance, she saw a log cabin and got excited. With instructions to place the bag of census materials on the door handle, she ventured forth to the front door. And there, already hanging on the door knob, was a census form… from the previous census, ten years earlier.
And now, some of my favourite points:
- It costs 42 cents for a mailed back census form. It costs $57 to obtain census information for every household that does not mail the form back.
- Every 1 percent increase in mail back rates saves 85 million dollars.
- PSAs for the census generated 800,000,000 impressions, with Karl Rove being the most important speaker in them.
- The US census website includes both positive and negative videos about the census.
- They created a K-12 curriculum targeted to 56 million students and included take home materials which were translated into 28 languages. Why? Because students are influential in their parent’s behaviour, particularly those new to the US.
- They do a test-retest design of a small portion of the data to examine the reliability of the information.
- They allow proxy surveying if a household cannot be reached after 6 contacts, e.g., neighbour or landlord.
- He admits that the census does burden people and interrupt their lives even though it’s in a small way. Those intrusions are for the good of everyone and you should participate.
- “marginalia” is being read via scanners. People use the margins to comment on all kinds of relevant and irrelevant things.
- Twitter and facebook will be gone in 2020 replaced with new fads. What!??!!
- They are releasing the American Community Survey results very soon NOT the 2010 Census data. Do not confuse them. Tell your friends. Retweet/buzz/share.
Naomi must have got a big kick out of this presentation. Unfortunately, no one checked Robert’s presentation to see if all of the colours would show on the screen.They didn’t. 🙂
Note: Any errors here reflect my lack of listening skills.
Conversition Strategies Social Media Research: By researchers, For researchers
- 2010 Census Comes in 22 Percent Under Budget (politics.usnews.com)
- Webinar – 2010 Census: Preparations for a Complete Count (kauffman.org)
- Care about State Coffers? Support the Census! (kauffman.org)