A correct graph and a misleading graph can be built from the very same numbers. The difference is choices — which shape, and where the axis starts.
Trap One — the Wrong Shape for the Data
The first big call is what kind of data you've got. Categorical data is labels with no order and nothing in between — favourite sport, eye colour. Numerical data is actual numbers, often flowing along, like a temperature measured each hour. Categorical data wants columns with gaps, because the categories are separate. Numerical-over-time data wants a line, because the line shows the trend between readings. Put a line on favourite sport and you've drawn a slope between "footy" and "netball" — as if there's something halfway between two sports. There isn't. That line is nonsense.
Trap Two — the Axis That Doesn't Start at Zero
This is the classic con, and you'll see it in ads and dodgy headlines forever. In the toy, three brands score 98, 100 and 103 — practically a tie. Start the y-axis at 0 and the bars look almost identical, which is the honest picture. Now slide the axis up so it starts at 95: suddenly Brand C towers over Brand A, even though it's only about 5% higher. Nothing in the data changed — only the bottom of the axis got chopped off, stretching a tiny gap into a cliff.
How to Stay Un-fooled
Two quick habits. First, ask "what kind of data is this?" before choosing a graph — that kills trap one. Second, whenever you meet a bar chart, glance at the very bottom of the axis: if it doesn't start at zero, mentally shrink the differences back down. The numbers on the axis are the truth; the heights are just the picture, and a chopped axis paints a lie.