Choose between the mean and the median, and say what each one hides.
Many numbers you meet in daily life are labelled "average": average salary, average house price, average commission. It is natural to read that as "what someone like me can expect", and often it isn't. In everyday speech "average" usually means the mean, and with some kinds of data the mean sits well away from what most people actually get. A claim can be accurate and still tell a newcomer almost nothing about their likely outcome. To read these numbers properly, you need to know the two common kinds of average, when they split apart, and what to ask before you trust a single figure.
The mean is the one most people learned at school: add all the values and divide by the number of values. The median is the middle value when you sort the values from smallest to largest. If there is an even number of values, there is no single middle one, so the median is halfway between the two middle values.
When the values are similar, the two measures come out close. Take ten colleagues with similar commutes. The mean and the median of their travel times will be near each other, and the difference rarely matters. The difference matters when a few values are far from the rest. The mean uses every value, so one extreme value moves it a lot. The median depends only on which value sits in the middle, so extreme values do not pull it.
Priya sees a recruitment post in their LinkedIn feed claiming agents earn an average of S$12,000 a month (an invented figure). Priya's cousin joined a sales team like that last year and earns nowhere near S$12,000. Both the post and the cousin's experience can be true, and the word "average" is the reason.
Suppose the team has ten agents with these monthly earnings: S$2,000, S$2,500, S$3,000, S$3,000, S$3,500, S$4,000, S$4,500, S$5,000, S$6,000, and the top agent at S$86,500.
To find the mean, add them up. The team total is S$120,000. Divide by 10 and the mean is S$12,000, so the post is accurate. To find the median, sort the values, which here are already in order. With ten values there are two in the middle: the fifth and sixth agents, at S$3,500 and S$4,000. Halfway between them is S$3,750, and that is the median.
Nine of the ten agents earn less than the "average". The top agent alone pulls the team mean up to S$12,000. Take that one person out and the mean of the other nine drops from S$12,000 to about S$3,722, while the median of the nine is S$3,500. One person moved the mean by more than S$8,000, and the median barely moved.
Neither measure is wrong. They answer different questions. The mean is the total shared out equally among everyone, which is useful if you are, for example, a company planning a payroll budget. The median shows what a typical person in the group sees. If you are one of the people in the group, or about to join it, the median is usually what you want.
A more useful thing for Priya, or any new joiner, to ask the recruiter is what an agent in their first year usually earns, and what the range is. The answer gets closer to Priya's own likely outcome than any single average can.
Statisticians call data like this skewed data: data that is naturally lopsided, with a long tail on one side. Income is the classic case. Most people earn within a narrow band, a few earn many times more, and those few pull the mean above what most people take home. For this reason, official income statistics are usually reported as medians. Singapore's Ministry of Manpower (MOM) leads with the median when it reports monthly income from work.
Property prices are skewed in the same way. Picture a condo development where seven units sold in a year. Six went for between S$1.15 million and S$1.38 million, and one penthouse sold for S$4.8 million. The mean price is about S$1.77 million and the median is S$1.3 million. A headline saying "average price at the development hits S$1.77 million" (also invented) would be technically correct, yet it describes none of the actual sales.
When you look up what a job pays, useful sources include MOM's Occupational Wage Tables, MyCareersFuture and a recruiter's salary guide. Check which measure each one uses. More generally, when "average" is attached to income, prices, commissions, wealth, or anything where a few cases can be very large, ask which average it is. If the source does not say, assume it could be the mean and treat it with care.
Even the median only describes the middle. Two teams can share the same median salary and look completely different. In one, everyone is clustered within a few hundred dollars of each other. In the other, half earn far below the median and half earn far above it.
What separates those two teams is their spread, meaning how far the values sit from the middle. A single number cannot show it. You don't need technical measures to ask about it. Ask what the lowest and highest values are, and what the bottom quarter and the top quarter look like. Some salary sources show a range around the median, such as the 25th and 75th percentiles. Those two figures mark where the middle half of people fall, so you can see how tightly or loosely pay is grouped.
The same habit helps at work, where an average often hides a few unusual cases:
Average deal size pulled up by one huge client. Average response time dragged out by a handful of tickets that sat for weeks. Average time to hire hiding that most roles fill quickly while two senior roles took a year.
When someone reports a figure like these, ask for the median alongside the mean, and look at the extreme values separately. The extremes are often the most interesting part of the data, because they point to the client, the tickets or the roles worth looking into.
The quickest way to feel this is to build the example yourself. In the activity below, you'll make up a team of ten with one very high earner and see how far apart the mean and median land.
Make up salaries for a team of ten, including one very high earner. Work out the mean and the median and compare them.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).