1. According to statisticians, variations from the average are more relevant than the average itself. For example, at an amusement park, waiting in line for a rise in the hot sun thinking, ‘if they just had one or two more roller coasters, all the lines would be shorter?’ It seems that adding attractions would lead to quicker queues, it’s actually not the case. Why?

It’s the varying patter of when guests arrive at the amusement park, and not the average number of guests arriving, that makes the lines so mind-numbingly long. Even if Disney could accurately predict the number of park visitors hopping on the Dumbo rise on a peak day, a line would still form because guests come at irregular intervals throughout the day, and because the ride’s capacity remains fixed. So, planning for capacity can deal with average rises in demand, but not with fluctuating demand. Disney has come to deal with this issue using a feature called FastPass, which entitles guests to come back to a ride at a designated time and jump into an express lane. FastPass works because it reduces the variability of guests arriving at any given time.

This is similar to traffic jams. Highway congestion is the result of suddenly high volumes of cars, which exceed the road’s average capacity. To deal with this, a strategy called ‘ramp metering’ can be adopted in which traffic lights on ramps regulate the pace at which cars enter the highway, thereby stabilising the number of cars on the road.

2.Statistics allows patterns being identified. For example, by doing in this way, credit modelers can determine a person’s creditworthiness. For example, in America today, many people take out mortgages quickly without undergoing invasive interviews. Creditors don’t need to ask people loads of questions before lending them money because a computer essentially already has, tracking a person’s innumerable past loan decisions and identifying patterns, or correlations. Using this information, a mortgage lender can instantly determine the creditworthiness of an application.

3. Decisions based on statistics face a trade-off between two types of errors. For example, if testing for athlete drug use, two types of errors can cause problems: the false positive, when an athlete who did not take drugs is falsely accused of doing so, and the false negative, when an athlete who did cheat is exonerated. As a result, the people administering drug tests face a trade-off: if they try to minimise false positives, an error that serves to diminish their authority, their actions will bear the unintended side effect of letting more cheaters walk free.

Lie detector tests are prone to the same unavoidable trade-off. Polygraphs work by monitoring the test subject;s medical statistics-like their breathing or blood pressure. The collected data can point to false information by showing a correlation with changes in the person’s physiological state. The detectives want to avoid false negatives, that is, criminals who go free. However, avoiding false negatives means accruing false positives and accusing many innocent people.

4. Statistics teach us to question patterns that appear obvious. For instance, on October 31st, 1999, an EgyptAir jetliner plummeted into the Atlantic Ocean off the coast of Nantucket Island, Massachusetts, leaving no survivors. But between 1996 and 1999, before this tragedy occurred, three other jets had crashed into the ocean right around the same location. As a result of the EgyptAir crash, some people stopped flying in the area. Their logic was that four crashes in four years was too many to be a coincidence, and that it must indicate some kind of disaster-prone pattern in the region. However, statisticians looked at the bigger picture, and while they saw the four crashes in four years, they also saw millions of planes safely traversing the same airspace during the same period. Statisticians approximate the odds of dying in a plane crash are one in ten million-about the same chances of winning the lottery.

Statistics are based on five key principles: search for variations from the average, uncover causation and reveal correlation, account for group differences, are realistic about the unavoidable trade-offs they face and question patterns whether they appear obvious or odd.