Statistics is a strange subject. Its purpose is to provide insights into many subjects. Yet, it provides only confusion. From classical statistics to Bayesian methods, the constant contrasts confuse so many. But, it shows just how beautiful our world is. It illustrates the world as part deterministic and part random. We model our finances, behaviour and simple tasks according to a set of statistical principles. Any event occurs with a probability and this probability is assumed to stem from some distribution. With many years of data the true data distribution reveals itself, but even with less observations we are able to assume this distribution and run many experiments.

But, it is also very frightening. Government policies are often made based on these statistical outcomes. Various surveys inform policy makers how much expenditure should be allocated to various population groups or to various states. Thus, not surprisingly, these surveys should conform to the strictest scientific rigour. This is unfortunately not always the case. The South African Statistics authority (Statistics South Africa or StatsSA) recently fired two of its Census statisticians on the basis of gross incompetence. This is based on results from a post census survey which measure how many people were not surveyed. The original results suggested something like 18% of the population (that is a massive number) which was changed to something like 14% after the Statistician General (SG) contended the original results. He said that he applied his logic and saw that the algebra did not make sense. Yet news papers only published the views of the Statistician General without getting the comments from the fired employees. There is seriously something fishy here:

• Jairo Arrow (the statistician originally in charge of the post census) has received many awards for outstanding service by the SG, but is now suddenly deemed incompetent
• The SG should be embarrassed for making premature statements regarding a single digit under-count
• The SG does not show how he applied his "logic" and "algebra" to the results   
• Experts at UCT also question the survey results, but this is met with laughter and a dismissal from the SG
• A 14% under-count is not very reliable to make informed policy decisions
• The SG handled the matter very poorly by quickly going to the press and blaming others for mistakes

Statistics in the hands of bullies is dangerous. To clear this mess up the Census methods and data should be made available for proper and professional scrutiny. The SG is quick to say that the Census results are good and a true reflection. I have my doubts however given the way things unraveled post Census.