In the future I predict Social Network Applications/Widgets could require Market Researchers to create entirely new sampling methods; creating a “cubed” or “Three-Dimensional” Sampling Plans.
The other day I was conducting cluster analysis on some of the most interesting data I’ve ever seen in my market research career! Big statement I know, especially considering the thousands […]
In the future I predict Social Network Applications/Widgets could require Market Researchers to create entirely new sampling methods; creating a “cubed” or “Three-Dimensional” Sampling Plans.
The other day I was conducting cluster analysis on some of the most interesting data I’ve ever seen in my market research career! Big statement I know, especially considering the thousands of projects, surveys, clients I’ve worked for, and then more recently data Anderson Analytics has collected from the web by screen scraping and text mining consumer generated media.
What I was looking at last week was data from one of our social media applications/widgets on Facebook. There was data on over 45,000 users. However, I limited the clustering to the 3,200 or so recent active users. Then fed the segmentation analysis back into the application using discriminant analysis to create a typing tool.
What’s so fascinating to me is that this really allows for creating a cube of data. What people say about themselves as well as what their contacts say about them. It’s like 3-D Market Research!
Anyway, I was pressed for time, but still wanted to improve this app a little since it was covered in AdAge this past week. If you want to try it out and see which of the 6 segments you fall into you can go to the application here: http://www.facebook.com/apps/application.php?id=43439829009
You can also add a badge to your profile page showing how you score for each of the 6 dimensions and see which of your friends are similar to you. In my badge you can see that I’m most like the “Rock Star” segment, i.e. I have a rather healthy self image 😉 I am most different from the “Tough Times” segment which I assume also makes sense; after all I can’t very well lay myself off, LOL.
Anyway, I think leveraging social networks like this is the coolest thing to happen to market research since, I dunno, Al Gore invented the web. This may require us to think of entirely new sampling methodologies. Forget a random sample described as “n=3,000 US Rep”. In the future we may be talking about something more like “N=48,000 ~ n=3,000 Rep + n=45,000 of their first level (family, friends and colleague) connections”.
Tom
Link to original postTom H. C. Anderson – Anderson Analytics
