Short answer: rolling a 6 on a fair die is P = 0.1667, so 0.8333 it does not happen and odds of 1 : 5. Two coins both showing heads is 0.25, and at least one head is 0.75.
Counting versus measuring
Probability from counts is exact and needs the assumption of equally likely outcomes. Probability from data — how often a bus is late — is measured, and the two must not be mixed: a bus that is late 3 days in 10 has a frequency of 0.3, not a probability of three tenths in any exact sense.
Worked example
A bag holds 4 red and 6 blue balls. Drawing red is 4 ÷ 10 = 0.4 and blue 0.6, with odds of 4 : 6. Drawing two reds without replacing the first is 0.4 × 3/9 = 0.1333 — and that second factor is why the page insists the independence assumption be checked before multiplying.