While reading a blog post for work I came across one of my favorite statistical distortions.
Look at this chart. Isn’t it amazing? Visually, it seems that yahoo mail users simply have way less friends than Gmail users! How sad for them. Maybe if they got Gmail they would be more popular.
Here is the same data formatted differently.
Looks like those yahoo mail people have friends after all. They have almost as many as most of the other e-mail users and only 6 less than the Gmail users. By changing the axis range we are able to completely alter the visual message of the chart. Therefore we can see that careful selection of scale and range is an extremely important part of presenting statistical data.
Another issue with this chart has to do with lumping all the user age groups together for the average. Since Gmail users tend to trend younger, it is possible the difference can be explained by demographics. It makes more sense to segment by age-group and then compare. I would like to know if there is any difference in the average friends per user between mail services in the 30-45 year age group. That would be more interesting to know.
To be a fair, the first chart is the default one you get when you enter the data into PowerPoint or Excel. So you have to be careful and edit the axis to avoid this problem.
You can find the whole article here: Link
On FriendFeed, this post was liked by 0 people and commented on 0 times hide
