Flip a penny and a dime. The penny can land H or T, and for each of those the dime can land H or T, so the full list is HH, HT, TH, TT, with the penny's result written first. These four are equally likely. "One head and one tail" covers two of them, HT and TH, so its probability is 2/4 = 1/2. Write the list as "no heads, one head, two heads" instead, and you get three results that are not equally likely: one head happens twice as often as either of the others.
For bigger experiments, organize the list so nothing slips through. Two dice make a 6-by-6 grid of 36 ordered pairs, from (1, 1) to (6, 6). A meal deal with 3 sandwiches and 4 drinks has 3×4=12 pairs. An event is then just a part of the sample space: "doubles" is the diagonal of the dice grid, 6 squares, so P(doubles) = 6/36 = 1/6.
A list of possible results is not automatically a sample space you can count. "Rain or no rain tomorrow" names two results, but they are not equally likely, so dividing 1 by 2 tells you nothing. Counting outcomes works when they come from fair coins, dice, cards or equal spinners; for weather or sports, you need each outcome's probability from data.
Think of it like this. A sample space is the class roster before a name is drawn from a hat: every student on it exactly once. Miss a name and everyone's chance is miscounted; write a name twice and that student's chance doubles. The roster works because each slip in the hat has the same chance; for lopsided outcomes, a list alone is not enough.
Core idea. Every probability is a count inside a complete list: write down every possible outcome exactly once, make sure the outcomes are equally likely, and an event's probability is the share of the list it covers.