Everyone, and especially modellers, should read Tim Harford’s book, How to Make the World Add Up. The aim of the book is to encourage people to understand statistical truths and to “escape the grip of the flawed logic, emotions and cognitive biases that shape the falsehoods”.
Harford’s top ten tips are as follows:
1. Search your feelings
We see what we want to see, for example farmers and bakers will give different forecasts for wheat prices. Try to notice your emotional feelings.
2. Ponder your personal experience
We need to use our common sense and our own experience. Spreadsheets and statistics cannot give us all the answers.
3. Once you know what the question is, you will know what the answer means
Don’t jump to conclusions without understanding the statistics. Be curious and ask the right questions.
4. Step back and enjoy the view
Retain a sense of proportion and put information into context.
5. Get the back story
There is a systematic bias towards publishing interesting results. The media often fail to give the back story and will report on scientific research without giving the full picture, including contradictory evidence.
6. Ask who is missing
Any assumptions in a model are the result of preconceptions and oversights. Statistics can give a very skewed impression of the truth if vast numbers of people are missing from the analysis.
7. Demand transparency when the computer says “no”
Like people, algorithms are sometimes trustworthy, sometimes not. We should be sceptical of both hype and hysteria and should not simply trust that algorithms and computers always do a better job than humans.
8. Don’t take statistical bedrock for granted
Countries need teams of skilled, professional statisticians. Their independence should not be taken for granted.
9. Remember that misinformation can be beautiful too
Check your emotional response to visuals and that you understand the basics behind any graphics. Just because a picture grabs your attention, it is not necessarily accurate.
10. Keep an open mind
Be prepared to change your opinion. Attributed to John Maynard Keynes: “When my information changes, I alter my conclusions. What do you do, sir?”
Most of these rules are of great relevance to financial modellers. We should also step back, think logically, never jump to conclusions and always remember that garbage in is garbage out.