I read the article and I am taken back to what I learned in my 4 year degree. You can make statistics prove what you want,, you just have to ask the right questions.
I do think most promotion systems evaluation program stink and that is because there is no effective trial period with most promotions. It is a one-and-done deal. Most places I have seen with a "trial" system either make the trial period too short (Is a year REALLY enough time to discover their faults or just short enough that they can BS through it?) or the trial parameters lack teeth to remedy the situation when a bad promotion happens.
I like the premise of the TV show "undercover Boss" where the CEO or head of a company goes to work "undercover" as a lower lever employee in their own organization. They then hear the real problems that plague the company.