I think the float will sit a little bit higher in water than in gasoline because it will not have to displace as much water to reach the point of equilibrium, so your gauge readings would be slightly higher with using water. I am guessing that the difference in the float depth is probably pretty small compared to the overall range of the float travel of the sending unit from full to empty.
Michael Driskell posted his gauge calibration process for a resistance stlye sending unit in the recent discussion about replacing the fuel tank on a 1947 car, post #14 in the attached link.
https://www.fordbarn.com/forum/showthread.php?t=287943