The only proper way to test a resistor is out of circuit. You've got to remove it before you can get a reliable resistance reading.

With power off, set meter to measure resistance.
If the multimeter has multiple range options, use the one appropriate for whatever value the resistor is supposed to have (you can tell from the colored bands on the resistor, and there are plenty of pages online that will do the calculation for you). If it doesn't, make sure the resistor you're measuring is within the meter's range.

1 probe on the lead at either end of the resistor, polarity doesn't matter. Make sure you're making good contact, because if you don't, your reading will be higher than it should be.

Resistance is measured in Ohms (omega symbol), and abbreviations K and M are used for thousand or million ohms (ex, 10 K Ohm = 10,000 ohms) - keep this in mind when setting range and reading resistor values.

This depends entirely upon the circuit. I can think of lots of examples of resistors that could be measured just fine in place, and lots of examples where they couldn't be. You need to disconnect one end of the resistor to be sure you have an accurate reading in all cases.

That said, you get some information from measuring it in place - with power off, measure the resistance with it in place:

If the value is HIGHER than the resistor is supposed to be (note that analog multimeters have an inverted scale, needle moves more with lower resistance), then the resistor is definitely bad.
No reaction on touching the multimeter probes to a resistor (assuming you have the range set appropriately), would also mean the resistor is bad (specifically, completely open, aka infinite resistance).

If the resistance is about right, the resistor is probably (but not definitely - would need to test with one end disconnected to be sure) good.

If the resistance is lower than it should be, you have learned nothing, and you need to test with one end disconnected to get any information.