[hot wire]---[switch]---[load]---.
|
[neutral wire]-------------------+
|
[person]
|
[ground]
Edit: Or this one...
[neutral wire]---[load]---[switch]---.
|
[hot wire]---------------------------+
|
[person]
|
[ground]
the parallel resistance is lower
And once you've turned the power off, test wit with a non-contact voltage detector. A brand name one that's actually been built to UL/CE standards and is subject to actual recalls/etc is under $30. It's a straightforward, pretty much foolproof way to make sure you don't get yourself shocked.
A multimeter is, in theory, a fine alternative but in practice is not. The non-contact detector detects the presence of the fluctuating magnetic field from A/C. A multimeter measures the difference in potential between two points. If you were to, say, have 120V on from same phase on both hot and neutral in an outlet, it would read 0V. If, say, the ground were also not connected then you'd also get 0V testing either against ground. By all accounts there is no voltage in the plug. If, say, each side of the plug were on a different breaker you might think you'd killed the power while it's still live. (Yes, this is based on a true story.) The only way to avoid this is to make sure you have a known good ground to test against. Whereas a non-contact voltage detector immediately lights up like a christmas tree telling you this situation is all sorts of fucked up.
Depends on how it is wired. Sometimes lights are wired so that the switch comes before the fixture, sometimes after. One of several reasons you should always turn power off at the breaker when fiddling with the wiring, instead of assuming it won't be hot if the switch is off.