a bias probe to stick in the socket is probably the safest way to check bias yourself. sometimes it will still require you to take the chassis out of the shell though, which carries a risk.
the way they work usually is thus...
They take a voltage from the anode of the valve, say 480volts DC (the plate voltage)
They measure the voltage present across a 1ohm resistor between the valves cathode and ground. The 1ohm resistor makes the maths easy to convert the values so 1mV = 1mA. The resistor is usually part of the bias meter. So a reading of 26mV means the idle current draw of your valve would be 26mA. You measure each valve to check one isn't drawing too much more or little current than the others, but matched sets should be reasonably close.
So then you end up with 480volts DC on the plate and an Idle Current of 26mA. Once you know that you can use something like this...
http://www.ax84.com/biascalc.html. Put in the plate voltage and the tick the valve type and you'll get the desired Idle Current for 60%, 70%, 80% etc disapation (disaption being the maximum output of the valve).
So in this example, say we used El34's... we'd want an Idle Current of around 43.75mA per valve @ 480volts. So
then you adjust the bias voltage itself (what the peavey test points tell you) which in turn will raise the idle current from 26mA up to 43.75mA (or 26mV / 43.75mV as displayed on the meter). This is all best rule of thumb. Some people use only their ear, some use an oscilloscope. As you can see from all that, knowing the negative bias voltage alone is only part of the story.
Anyway, that is an easy and reasonably safe way to go about it with the minimum of fuss and understanding. Sometimes the idle current includes the power drawn by the 'screen' of the power valve, which means you end up biasing a few mA colder than you think, this would depend on the design of the meter you use to test with.
I have a personal method for setting bias using only a meter, but I can only do that with one of my two meters (a more expensive true RMS fluke). It took me a while to get confident with that method since you need to poke around at some of the highest voltages in the amp, so it's not worth going into.
What Peavey probably do in the JSX (if its like how the 5150 is biased), is give a suggested voltage that actually biases the amp really cold. my old 6505 would idle at 14mA per valve, which is SUPER cold. I added a bias mod and adjusted it up to 36mA. It sounded louder and less fizzy. So much better. The advantage of a bias that cold is that you could pretty much put in any matched set and it would be accidently too hot whatever. Making maintainance a bit easier especially since on a stock amp like that you can't adjust the bias anyway.
