Ethical

Lunched with LS. She was, as often, not very happy.

“What’s the matter this time?” I smiled.

“It’s that hopeless study again,” she frowned, presuming that I knew which study she was talking about (and yes, I did – after listening to it for half a million times). She continued, “The data are chaotic, and the result just does not turn out to be the way I want. I’ve tried no fewer than half a dozen of methods to analyze the data – so much so I am coming to the conclusion that my statistics is not very ethical!”

“You know, it is a self-conflicting idea to have an ethical statistical analysis,” I said.

“Alas, don’t go that far, my friend,” our professor of rheumatology protested.

“But yes,” I became slightly more serious, “The situation is just like cookery. If you have a good fresh piece of fish, you often don’t really need to cook – you can just slice it and serve as sashimi; elaborated cooking with seasoning and spices, however, is always necessary when the fish at hand is half rotten. Similarly, if everything of the study is in good order, you can usually describe and present and forget about complicated statistics – which is absolutely useful when the data are problematic.”

“But, isn’t statistics a technique designed to help us see things more clearly?” My friend asked.

“Yes, it was,” I murmured, “But, only was. There used to be an old saying: Language can express what you think, hide what you think, or replace the need of thinking altogether. If you follow Wittgenstein, statistics is a sophisticated form of language; it can present your data, but, more often than not, serves to hide your data, and, looking at the research of some epidemiologists, sometimes replace all data!”