the Tech Research meet in the Clachan – and my 3 challenges

Clachan Last night I headed for the Clachan in Kingly Street for a gathering orchestrated by Mario Menti and Surindar Siama. Of those involved in the technical end of research – technical managers and software developers. High on my radar because I had never heard of such a group and had questions for them which I didn't think I could ask sensibly anywhere else.  There were a dozen people there including a rogue client from BA who was drawn for much the same reasons I was. And a great night was had by all. I promised to write up my 3 challenges so here they are.

1/ When can we have a survey engine for the iphone built using the free SDK – see my earlier blog on this. The iphone user base in this country has passed 3 million. All are using the cheap if not free downloads from the appstore so are familiar with the standardised interface. The small screen might force some discipline in survey design too. Its clunky to download surveys from a software delivery system but a run of such activity might encourage Apple to set up a research portal alongside appstore and itunes for downloading surveys. On the grounds that they are worth more in revenue terms than software apps. Surveymonkey I was told claims 2 million active account holders creating surveys. Wouldn't it be even better doing this on a mobile platform? Pat Molloy grinned at me in a predatory way and said this was an area he was already working on.

2/ Can someone build me a software sampling agent I can use to tag and assemble online samples? Most online surveys are like horseless carriages betraying their ofline origins. Here's the convention. Recruit a sample then fire questions at them. This is a all getting dull and the lakes are starting to dry up as respondents succumb to extreme cynicism or boredom. At the other extreme websites, blogs and twitter are throwing up all sorts of interesting content. And we can track those who post it. But we don't have the means to arrange them and weight their content as a sample. To make sure they are female, 25-35 and live in Western Englands and not LA or Mindanao. I would envisage a tagging system like delicious or digg which would tag demographic indicators – there also might be a basic survey which we could use for people to tell us a bit about themselves. Would we officially recruit them? No. Could this turn into an useful type of research. A thousand times yes and better than the scraps that scavengers bring in and use as the foundation for insights.  Mario Menti has started to scratch his head over this one.

3/ How can I use outliers to learn useful things about the core sample? Online research as Tom Ewing has memorably said requires the outliers to carry most of the weight.  80% of content comes from 20% of the people. Conventionally statisticians like to harvest the centre of the distribution curve throwing out any outliers because they throw the findings out of proportion. However if you don't have access to the bulk of the online population because they browse and don't post. Then is it possible to determine where they stand and what they think by studying the outliers instead? Sounds loopy because it flies in the face of current practice. But the only way online research is made acceptable is by pretending that those who choose to participate online are normal and representative when almost certainly they are not.  The other thing I am betting on is that the distance between the outliers or early adopters and the mean is quite stable so it should be possible to generate a coefficient for each outlier and tune this.

There was a lot of hand waving at this point and comments that this was impossible. I provided 2 counter examples. Firstly that artillery works entirely using outliers – you use your misses to work out how to hit the target.  And secondly that by studying Daily Mail Journalists, what they write and their presumed audience you can learn a lot about what Daily Mail readers think. Journalists aren't representative but they are useful outliers who we use all the time. Fraid I didn't get an answer to this one because there were no statisticians in the house.  But pass this on to a statistician near you and perhaps this challenge will keep them up at night.

Great fun – I hope they meet again soon. As I returned on the train I was reflecting that this was a typical piece of planning thinking. Many people think planning is about finding the right answer. Its not. It actually searching to find the most interesting and useful questions. I'll let you know what happens as a result and if you dear reader want to work on any of these then be my guest but do get in touch and tell me how you're getting on.



Twitter Updates
Photo Albums
Grindleford Grinch April 2007Putna Sihastra Monastery 2011Willie Williams exhibitionTV stations JeddahPlanners drinks night No1 Bucharest

Designed by Matthew Pattman