When a graph uses a scale (e.g., each square represents 5 people), the reader must multiply to find actual values. If a bar reaches '2' on a scale of 1:5, the actual count is 2 × 5 = 10. Misreading scales is a common error.
Before reading any value, identify the scale by examining the axis labels and the gap between them. Practice with graphs that use different scales (2, 5, 10) so scale-checking becomes automatic.
You already know how to read a bar graph where each square means 1. You've also worked with picture graphs that have a key — where one picture might stand for 2 or 5 real items. Scaled graphs are the same idea applied to any graph: instead of each unit on the axis representing 1, it represents some larger number. That number is the scale, and it tells you the multiplier.
The first habit to build is always reading the scale before you read any data. Look at the axis that has numbers on it — in most bar graphs, that's the vertical axis. Find two consecutive labeled values (say, 0 and 5, or 0 and 10) and note the difference. That difference is your scale factor. Every time you read a bar or picture value, you're reading how many "scale units" tall it is, and you multiply by the scale factor to get the real value.
For example, if a bar graph's y-axis goes 0, 5, 10, 15, 20 and a bar reaches up to the 3rd line, you're at 15 — not 3. If you forget the scale, you'd read 3, which is five times too small. The multiplication is always the same: graph position × scale factor = actual value. For picture graphs, the key does this for you: "each symbol = 5" means you count symbols and multiply by 5.
Scaled graphs are used because real-world data often involves big numbers that wouldn't fit on a graph with a scale of 1. A survey of 200 students couldn't be shown with 200 individual squares. A scale of 10 compresses the graph to 20 squares — much more practical. Understanding scales lets you read charts in newspapers, textbooks, and scientific reports, where authors always choose a scale appropriate to the size of their data.