Picture the class tipping out their recess: six apples, four bananas, three oranges, two bunches of grapes. That's a tiny dataset — four categories and a count for each. You could just leave it as a list, but a list is hard to feel. A graph turns those counts into something your eye reads in a second.
One Table, Three Pictures
Hit the buttons in the toy and you'll see the very same four numbers drawn three ways. A column graph gives each fruit a bar, and taller means more — apples stand head and shoulders above the grapes. A dot plot stacks one dot per piece of fruit, so you're literally counting the dots. A divided bar takes the whole lot — fifteen pieces — and chops one long bar into slices, which is brilliant when you care about parts of a whole: how big a slice is "apple" compared to everything brought in?
Why Bother Having Different Pictures?
Because different questions want different shapes. If you want to compare amounts — who got the most? — columns or dots win, because heights and stacks are easy to line up. If you want to see how the whole splits up — what fraction was apples? — the divided bar is unbeatable, because the whole bar is the total and each slice is its share. Same data, different job, different picture.
The Thing to Carry Forward
Once you believe a graph is just a re-drawn table, two things follow. Drawing one (rung 2) is easy — you're only copying counts into heights. And reading one (rung 4) is the same move backwards — trace a bar to the axis and you've recovered the number. Everything in this concept is that one idea, forwards and back.