One sample is not the whole story
If you take a different random sample, you get a different estimate. That is variation.
Here, 30 different random samples of 20 students were taken from the school. Each dot is the percent who ride the bus in one sample.
The dots run from 15% to 60%. But they pile up near 40%. The center of the dots is the best guess for the whole school.
Example
How many of the 30 samples gave an estimate from 30% to 50%? What does the plot say about the whole school?
Count the dots at 30, 35, 40, 45, and 50. There are 2 + 5 + 5 + 3 + 7 = 22 dots. So 22 of the 30 samples are in that range. The dots pile up near 40%. So I infer that about 40% of the school rides the bus. Check: 22 out of 30 is a big majority, and the real school has 80 out of 200, which is 40%.
Trap: A common mistake is trusting one sample too much. One random sample gave 15%. Another gave 60%. Both were fair samples. That is why we look at many samples.