All models are approximations. Essentially, all models are wrong, but some are useful. However, the approximate nature of the model must always be borne in mind.

Statistics is, or should be, about scientific investigation and how to do it better, but many statisticians believe it is a branch of mathematics. Now I agree that the physicist, the chemist, the engineer, and the statistician can never know too much mathematics, but their objectives should be better physics, better chemistry, better engineering, and in the case of statistics, better scientific investigation. Whether in any given study this implies more or less mathematics is incidental.

Statisticians, like artists, have the bad habit of falling in love with their models.

All models are wrong, but some are useful.

Management must provide employees with tools that will enable them to do their jobs better, and with encouragement to use these tools. In particular, they must collect data.

The old-fashioned idea of a good manager is one who is supposed to know all the answers, can solve every problem himself, and can give appropriate orders to his subordinates to carry out his plans... A good modern manager is like a good coach who leads and encourages his team in never-ending quality improvement.

All models are approximations. Essentially, all models are wrong, but some are useful. However, the approximate nature of the model must always be borne in mind.