Yes, you'll get a higher boil-off rate when doing a small amount in a big kettle
Another way of looking at this to not think in terms of rate and percent, but in volume/hour. For example, if you are boiling with the same flame on the same pot, you might be losing 1 g/hr, regardless of your volume (since you've got the same flame and same kettle geometry). If you were to only have 2 gallons in your kettle, you would have a different rate (not 9% per hour), resulting in a higher percent being lost per hour.
I'm assuming it will also vary based on the shape of your kettle (tuna can vs soda can).
It likely doesn't scale just based on volume/hour either... perhaps some sort of logarithmic type relationship? If I get ambitious I may play around with measurements, but the long and short of it is that boil off loss (as measured by % lost) is going to vary based on your starting volume. The rates put out there are based on a standard size batch, so when JZ for example gives his rate loss, it is based on the standard Jamil Show preboil batch size.