How percentages change when the base changes

A percentage only has meaning relative to its base, the starting amount it is measured against. The same absolute change can look large or small depending on that base, and the same percentage applied to different bases produces different absolute changes. Before you judge any percentage, ask percent of what, and then ask how large the starting amount is. Those two questions answer most of the confusion.

Here is a pair of numbers that look different but describe the same kind of change. A group of two people grows to three. That is an increase of one person, which is 50 percent of the original two. A group of 200 grows to 201. That is also an increase of one person, but only 0.5 percent of the original 200. The absolute change is identical, while the percentages differ by a factor of one hundred.

Same change, different bases

Percentage change is the difference divided by the starting value, multiplied by 100. A rise from 40 to 50 is 10 divided by 40, which is 25 percent. A rise of the same 10 units from 400 to 410 is 2.5 percent. The arithmetic is simple, but in conversation the base is often missing. A headline that says a figure rose 25 percent tells you little until you know whether it started at 40 or at 400.

The same logic works in the other direction. A fixed percentage applied to a large base produces a large absolute change, and applied to a small base it produces a small one. If a figure grows by 10 percent in each of two years, the increase in the second year is larger than in the first, because the base has grown. A steady percentage can mean a rising amount, which is why a percentage alone can hide the scale of what is happening.

The table applies the same 10 percent change to three starting values. The percentage never varies, but the absolute change grows tenfold each time the base does.

Scroll sideways to see the full table →

Also remember that a percentage compares two figures only when both come from the same base. A share of a total can move between two years even when the underlying count stays the same, because the denominator moved. Check whether the base is the same in both figures before reading the difference as a real change.

Small bases produce dramatic percentages

When the starting count is small, a single unit can produce a huge percentage. Going from one case to four is an increase of three cases, which is 300 percent of the original. Going from one case to two is 100 percent, even though the change is still just one case. Dramatic percentages from small counts are common in local figures and in small groups, and they often reflect the small base more than a real change in the underlying situation.

Before reacting to a percentage, ask for the starting count. If the report gives the base, you can judge the change directly. If it does not, a dramatic percentage with no base should prompt a search for the absolute numbers rather than being taken as evidence of a large effect. The absolute numbers may show a change of one or two cases, which is a very different story.

The same effect appears inside subgroups. If a subgroup of 25 respondents gives an answer of 40 percent, that is 10 people. If one more person chooses that answer, the share moves to 44 percent, a change of 4 percentage points produced by a single response. Subgroup results deserve the same base check as headline figures.

Why a rise and a fall do not cancel

Percentage changes do not reverse neatly. Suppose a value of 100 falls by 50 percent to 50, and then rises by 50 percent. The rise is 50 percent of 50, which is 25, so the value returns to 75, not to 100. The overall result is a 25 percent loss. The second change was measured on a smaller base, so the same percentage moved less in absolute terms.

The same thing happens in the other order. A 50 percent rise followed by a 50 percent fall takes 100 to 150 and then to 75, again a 25 percent loss. Reversing the order of two changes gives the same final result, because multiplying 0.5 by 1.5 and 1.5 by 0.5 yields the same product.

The effect appears with smaller moves too. A 10 percent rise followed by a 10 percent fall leaves the value at 99, not 100. Checking a sequence of percentage changes is easiest if you convert each change into a multiplier, such as 1.10 for a 10 percent rise and 0.90 for a 10 percent fall, and then multiply the multipliers together. The product gives the final value relative to the start, and it makes the lost ground visible.

When you read a percentage, find the base, find the absolute number behind it, and check whether a sequence of changes was measured on a shifting base. Those steps take seconds. They show whether a percentage describes a large movement or a small one measured against a small starting point. For decisions, the absolute figures usually say more than the percentage does.