"Sample size of 1 office but it's enough for me to steer clear of their hardware."
That doesn't make a lot of sense considering minimal research would have revealed that this is not representative of the situation overall.
So, according to the decision model you are using, if you saw your friend win the lottery, you'd start playing every day and expect to win, since the statistical likelihood of something happening is less reliable to you than your anecdotal experience.
http://cache.consumerist.com/assets/resources/2006/10/laptop...
My anecdotal experience is also that macs are a little bit unreliable, but this is obviously a subject on which we should trust the data (which says that macs are slightly above average.)
According to the context ( http://consumerist.com/210708/all-laptops-break-period ), this is apparently a graph from 2006 ranking "how quickly different laptops need repair," which doesn't really make a lot of sense given the way the graph is displayed. These definitely aren't the reliability scores from the 2006 rankings, which you can see discussed here: http://arstechnica.com/apple/news/2006/10/5614.ars
FWIW, the upcoming consumer reports rankings were announced yesterday, and all apple laptops topped the overall rankings:
http://apple20.blogs.fortune.cnn.com/2009/05/05/consumer-rep...
It's unclear where they are in the specific reliability rankings.