If the fitting is wired correctly, it doesn't matter much if the lamp is in place or not as regards measuring the voltage. Yes, you do connect a meter in parallel with a load (i.e. the lamp in your case) to measure voltage. But because a lamp doesn't take much current, it doesn't drop voltage significantly. Now if the load was high powered e.g. a heater, the voltage would drop a few volts. The open circuit voltage of a voltage source is always higher than the output voltage on load because a real voltage source always has internal resistance, plus the connecting wires have resistance also. So if the connecting wires are long or cross-sectional area is small, the voltage drop can be considerable if the wiring is sized inappropriately. If you connect the meter to the fitting without the lamp, it's in parallel with the output terminals on the fitting and because it's set to "volts", no current flows through it (well actually just a little, but microamps because it has such a high resistance). If the meter was set to "amps" it would be like a short circuit and effectively in series with the supply and a fuse would blow. Maybe the concept of parallel and series is a bit confusing. Just remember that when the meter is set to volts, it measures the voltage between two points and when set to amps, it measures the current flowing between the two points.