105 soldiers were participating in a 15 week commander training program in Israel. A psychologist informed the training officers who would be leading the program that they had assembled comprehensive information about each of the soldiers. Each soldier had been classified into one of 3 potential “Command Potential” categories: 1.) High Potential, 2.) Regular Potential, or 3.) Unknown – due to insufficient information. The soldiers were placed into these categories through a series of psychological tests, sociometric data from the previous course, and previous officer ratings. The training officers were requested to learn the names of each soldier who would be in his command and his “CP” rating before the program started. When the program began, the soldiers were divided into groups with even numbers of high, regular, and unknown CP soldiers.
After the 15 week program ended, the officers confirmed that those soldiers who were placed in the highest “CP” category were in fact the ones who had performed the best.
Here’s the catch. The “CP” (command potential) ratings were not actually assigned based upon any data at all. They were actually selected at random.
Those soldiers who were seen as having high potential ultimately lived up to the expectations of their commanding officers. When the officers were told about the random selections, they argued with the psychologists and defended their 
position – they truly believed that there must have been actual data used to reach those conclusions.
Expectations matter.
I wonder how many of the “regular CP” or “unknown CP” soldiers actually had a lot of potential too? I wonder what effect lower expectations have upon someone?
PS – This example was also from the book “Sway” by Ori and Rom Brafman. If you like this kind of stuff, you should check it out.