Good question, it's a very confusing subject! For a "ratio", the null value is 1 but for a "difference" the null value is 0. So it depends on if the measure of association is looking to find a difference between two groups or if it's trying to divide them and create a ratio. If you think about it conceptually, when you're trying to find if X is significantly different (greater or less) than Y, 0 would indicate there's no difference. However, if you take a ratio of X/Y to compare (for example a risk ratio), then the math doesn't work out for 0 to be a possibility. For example, if X is 1 and Y is 1 (X/Y = 1) and the ratio is 1, which means X and Y are essentially the same. So just always ask yourself when thinking about a null value, "what is the measure they're trying to establish" and once you decide if its a ratio vs. a difference you know if its 1 vs. 0