Reading a value from a graph within the range of your data is called interpolation, and it is a common requirement in data-based questions. The key principle is to read from the best-fit line, not from an individual plotted point, because the line represents the overall trend and averages out scatter.
Start from the known value on one axis, draw a light guide line straight across or up until it meets the best-fit line, then draw a second guide line to the other axis. Read the value where that line meets the scale.
Accuracy depends on careful reading of the scale. Count the small squares and estimate sensibly between marks rather than rounding to the nearest large gridline.
Because interpolation stays inside the measured range, it is more reliable than extrapolation, but it still depends on a well-drawn line and precise reading, so take the same care you would when plotting.
What examiners award
Marks are awarded for reading from the best-fit line using guide lines to the axes and reading the scale accurately.
Common mark losses
In short
To read a value within the data range, use the best-fit line, not the scattered points. Draw guide lines to the axes and read carefully off the scale.
A routine that works
Practise the mechanics until they are automatic: label both axes with units, choose a scale that fills more than half the grid, plot points with a sharp pencil, draw a single best-fit line, and take the gradient from a large triangle with its coordinates and unit shown. One clean example beats re-reading the theory.
Source: DSKP KSSM Physics Form 4 and 5 (Versi English) (Bahagian Pembangunan Kurikulum (BPK), KPM)