Sunday, October 28, 2007
Better than Six
Those 6 elements that constitute on-line research quality
approach to build
management
sampling
activity and qre design
cleaning and editing
system reporting
That's what innovation looks like....and it is what our clients need to see. It is externally focused. Let's go there.
approach to build
management
sampling
activity and qre design
cleaning and editing
system reporting
That's what innovation looks like....and it is what our clients need to see. It is externally focused. Let's go there.
Monday, October 22, 2007
Data....But Not as We Know It
....Yes, you know the stuff - numbers collected from surveys. Fall in it. Live it. Love it.
I am interested in data for two reasons at present:
a) metadata: using background passively collected data to add insight power
b) number crunching. Using historical data sets to be able to make better decisions going forwards - of course Amazon referrals is a classic example. www.farecast is another brilliant example. Research wants foresight....
Two research buzz words this taps into:
- crowdsourcing
- insights
I am interested in data for two reasons at present:
a) metadata: using background passively collected data to add insight power
b) number crunching. Using historical data sets to be able to make better decisions going forwards - of course Amazon referrals is a classic example. www.farecast is another brilliant example. Research wants foresight....
Two research buzz words this taps into:
- crowdsourcing
- insights
Sunday, October 21, 2007
Data Warehousing
Recently I have been looking a great deal at data warehousing....
If we think about Netflix, Google, Amazon and the like what they do really well is to harness collaborative filtering or crowdsource to use the right lexicon. This builds upon Weinbergers ideas of the Third Order of Order where we can expand our creativity and intellectual power through the ability to cut data streams in new ways andhence extract new meaning and value from events, data and information.
Can we do this with research? Research always follows marketing.
If we think about Netflix, Google, Amazon and the like what they do really well is to harness collaborative filtering or crowdsource to use the right lexicon. This builds upon Weinbergers ideas of the Third Order of Order where we can expand our creativity and intellectual power through the ability to cut data streams in new ways andhence extract new meaning and value from events, data and information.
Can we do this with research? Research always follows marketing.
SIXySampling
When it comes to sampling we need to think about a number of things. And perhaps there are two things which are not often discussed that go beyond quotas, randomisation, databases, profiling, sample and all of the usual jargon....
- outsource de-duplication to ensure the same panelist doesn't complete the same survey from two different sample sources
- some balancing of the sample from the database. Without this it is difficult to make results replicable and robust
- outsource de-duplication to ensure the same panelist doesn't complete the same survey from two different sample sources
- some balancing of the sample from the database. Without this it is difficult to make results replicable and robust
Monday, October 15, 2007
SIXteen ten / ten SIXteen
Panel Quality equals research quality? Is that 100% true?
Is all (on-line) research panel research?
If you provide a long questionnaire or an inadequate questionnaire does your panel quality impact the quality of the end data in it's entirety? Do we need to revise our thinking to adopt this, not just say 'yes'?
Can the layout of the questionnaire impact the end data as much as the panel itself?
Does this way of thinking require organisational and commercial adaptations?
Is all (on-line) research panel research?
If you provide a long questionnaire or an inadequate questionnaire does your panel quality impact the quality of the end data in it's entirety? Do we need to revise our thinking to adopt this, not just say 'yes'?
Can the layout of the questionnaire impact the end data as much as the panel itself?
Does this way of thinking require organisational and commercial adaptations?
Tuesday, October 09, 2007
Four And Two
Welcome to the second release.
Even with the control mechanisms on a panel the fact is that the end data is not always perfect. key lesson - great panel does not equal great data. Still with me?
Many of us have mechanisms to remove people who complete surveys too quickly. Some of us enforce it. The fact is that an automated mechanism is good but artificial intelligence cannot pick up all of the obvious poor data provided (for whatever reason). So we need the human visual edit - wow there go some of my efficiencies!
See you soon....
Even with the control mechanisms on a panel the fact is that the end data is not always perfect. key lesson - great panel does not equal great data. Still with me?
Many of us have mechanisms to remove people who complete surveys too quickly. Some of us enforce it. The fact is that an automated mechanism is good but artificial intelligence cannot pick up all of the obvious poor data provided (for whatever reason). So we need the human visual edit - wow there go some of my efficiencies!
See you soon....
