This blog is about software engineering, services, graphics, soft sills, work and education. Feel free to rent a ninja for any task concerning these topics. Just leave a comment in case you are interested.

Monday, May 17, 2010

Visualize not comparable data

Over the weekend I was called to help at presenting information. The data was a survey concerning social networks and their marketing potential. As many surveys, questions like "how important is ... for ..." where asked and the answers where don't know, not, little and very. So, how to present this data? My client didn't want the usual bar chart you should use for that as the only info you've got is a number of votes. The data will be presented in a marketing paper so I can understand the need for fanciness and partly wrong usage of data. So I had to come up with some bubble chart. As I only have the amount of votes and a bubble chart needs at least three numerical attributes, I was able to convince my client to use a spider chart. The last problem was that the answer possibilities are hard to compare. What's very important compared to not important? A better question would have been, how much percent of your marketing budget are you willing to spend on this and that? That's an open question with a numerical value you can easily compare. So I have to convert the votes for options to a comparable single numerical value. The easiest way is weighting. Give three points to very important, zero to not important and so on. Depending on what you want to say, you can change the outcome of the diagram by using other weights. Like I could ignore the don't know answers or I can punish them with a negative weight. So by adding up the weight, calculating a dummy numerical value I came up with the chart presented in this post. It's a perfect marketing chart as it looks nice and doesn't provide any usable information (except for marketing people). If you want some charts too, contact me.

0 comments:

Post a Comment