Lies, Damn Lies and Statistics 2

Continued from Lies, Damn Lies and Statistics
One rule to always remember is "Just because is sounds true, doesn't mean it's true". If choosing a solution mean relying on statistics, keep the statistics simple. A statistic that is in a simple sentence can be analysed much quicker than if we were to question it's validity by involving the data. For example: If  that statement is "The device fails after an average of 14 months", it could be broken down into
  1. What does a failure mean? A problem that result in a replacement or device stops functioning requiring manual restart?
  2. Who did the test? The manufacturer or real-world experience by the person making the claim
  3. How long does it take to replace or fix? How long is the downtime? This takes a look at the affect of the statistics instead of the statistics itself. By assessing the statistics effect on choosing the solution, we can figure out how important it is and how important it be true. 

Keep an eye on the clock. Efforts to validate information may take too much time when there is little. As time ticks away, making a decision can become more important than making sure it is the right one.
    Know thy enemy is a strategy I strongly recommend. So in that spirit, I recommend the book "How to Lie With Statistics" by Darell Huff. As the title suggest, it talks about ways statistics can be manipulated. He gives real-life examples often submitted by other statisticians, depending on the edition you get hold of.  By knowing how statistics can be manipulated, you can spot potential problems with statistics you are reading. Asking the hard questions and keeping a healthy dose of skepticism will prepare you in facing lies being passed as statistics. 
    It is a popular book and was first published in the 50s. Just think of how many people read the book to learn how to lie with their data, statistics and analysis.