In preferences (at least in mathematica 12.0+), there is an option called “Global Magnification”. At first glance, this option seems to scale the size of the text in notebooks (e.g. 75% makes things smaller, 115% makes things bigger). But it in fact does quite a bit more. It will affect the dimensions of plots when using options like ImageSize (I learned this in my last question). At 75% Global Magnification, ImageSize->240 produces a 2.5 inch image. At 100% Global Magnification, ImageSize->240 produces a 3.33 inch image. Note, this later case is what you would expect from the documentation of ImageSize. The documentation states ImageSize works at 72 dots per inch. So 240 dots / (72 dots per inch) = 3.33 inches. It does note that 72 dots per inch is before magnification. So the different size images does make sense.
Further reading shows setting the default text size in mathematica is a different procedure. But a layman (me) probably wouldn’t know better.
So my question is what is Global Magnification really doing? What else does it change? And why might it be implemented in this way?