If the age-in population has a 75% incidence and the age-out population has a 25% incidence it won’t cancel out. The age-in population will be adding a lot more than the age-out will be subtracting.
Why would you lean towards assuming it differs by such an astounding amount over a relatively short time period?
Pretty much, there seems to be a natural gradient pre-existing, where the age-in group seems to have a much greater natural affinity to be living with their parents even prior to the pandemic. We can all probably generate reasonable hypotheses for that, but that's the root of the mathematical assumptions.
[0] Here's a re-link of what I'm referencing: https://www.pewresearch.org/fact-tank/2020/09/04/a-majority-...