Update 2011-04-28 16:32:39: I had a embarrasing numeric error in my post which I have now corrected. The result however, is in principal the same. My main grudge agains the hunch.com infographic was the wording used in the comparison.
A few days ago (2011-04-21), hunch.com published an infographic comparing PC and Mac users. It is one of the worst infographics I have ever seen. Why? Because the statistics is present say absolutely nothing and seem to be chosen based on troll factor alone! For example, take this comparision:
PC people are 36% more likely than Mac people to be late adopters. 43% of Mac people consider themselves early adopers.
I mean, WTF!
Fail #1: Not providing information on how many alternatives were provided. Does 43% early adopters among Mac users mean 67% 57% late adopters? Or were there additional alternatives – “somewhat early adopters”, “somewhat late adopters”. This could mean that there are 0% late adopters among Mac users.
Fail #2: Assuming that there were two choices, 43% early adopters mean, 67% 57% late adopters among the Mac sample. If PC people are 36% more likely than Mac people to be late adopters. As this infographic seems to be rather informal, this could mean several things. For example that
- 36% more PC people than Mac people responded that they were late adopters.
- The percentage of PC people considering themselves late adopters is 36% more then the percentage of the Mac people considering themselves late adopters.
- The probability of a PC user being a late adopter is 36% higher than the probability than that a Mac user is a late adopter.
In the first case, it would mean that If 100 Mac people answered that they were late adopters, 136 PC people would have answered that they were late adopters. However, the infographic states that 388,315 users participated and of them 52% were PC people, 25% were Mac people, and 23% were neither. In absolute numbers we therefore have 201,923.8 PC people and 97,078.75 Mac people. Using the information above, and the assumptions above, this means that there are 67% 57% of 97,078.75 late adopters among the Mac people, i.e. 65,042.7625 55,334.8875 late adopters among the Mac people.
Using the number of late adopting Mac people, we can calculate that 36% more late adopting PC people means that 88,458.157 75,255.447 PC people are late adopters. If 88,458.157 75,255.447 PC people are late adopters, 113465,643 126,668.353 PC people are early adopters. This means that of the PC people, 56.19% 62.73% are early adopters and 43.81% 37.27% are late adopters.
In the second case, it would mean that if we had 100 PC people and 100 Mac people, 43 Mac people would be early adopters. This gives us that 67 57 Mac people would be late adopters. If there are 36% more late adopters among PC people than among Mac people, this means that there are 24.79 20.52 more late adopters among the PC people. This means that of the PC people, 8.21% 22.48% are early adopters and 91.79% 77.52% are late adopters.
The third case is actually technically the same case as third case.
For both the second and third case, the wording regarding the PC people one of the least informative ones that you can use in statistics since it does not have any anchor point. 36% more late adopters among PC people could mean that if there were only 100 late adopters among the near 100,000 Mac people, 136 late adopters among the more than 200,000 PC people would still be 36% more than the number of the late Mac people. The actual percentages of late adopters would be 0.07% among the PC people and 0.1% among the the Mac people.
The kind of comparison made in the example (“PC people are 36% more likely than Mac people to be late adopters. 43% of Mac people consider themselves early adopers.“) has been used in several other cases.
There are more statistics presented in the infographic that are equally, or even more uninformational, but I’ll leave them as an exercise for the reader. However, if you do write about them, please comment below with a link, so I can point other readers to your posts. Also, if somebody from hunch.com
could tell me what the true numbers for the question above is, it would make me a little bit happier.