Statistics seem, almost by their very nature, to convey a positivist message. They are, in fact, a formidable tool in the attempt to master the complexity of the real world... But numerous "biases" threaten any discourse that refers to them without care: an over-simplistic reading of the figures can lead us - for example - to confuse correlation with causation... And more complex phenomena (notably Simpson's paradox) can distort conclusions that seem objective.
