A cuppla things--
If you have an unbalanced signal source and a choice to connect it to either a balanced or an unbalanced input, where the sensitivity and overload levels are equal in either case--then it's preferable to use the balanced input. Its common mode signal rejection will give you better immunity to noise from interference in the cable, PROVIDED that you use balanced cabling to make the connection.
Many people don't realize that a balanced input can give you noise immunity even if your signal source is unbalanced. But in order to get it, you MUST run a balanced cable from the signal source. To create this condition, find out the source impedance (output impedance) of the unbalanced device that's providing the signal, and then in your adapter from unbalanced to balanced (which MUST be at the "driving" end of the cable), provide that same impedance as precisely as you can between ground and the "cold" wire (the signal wire that you're not connecting to the output of your unbalanced device). Voila--much better noise immunity.
Secondly, the impedances of a symmetrical resistive pad become, in effect, part of the balanced input circuit of the preamp / recorder / mixer that you're coupling your signal into. It is for that reason that you want to match the values of the two series resistors in each pad to each other as closely as you can. Otherwise you throw off the balance of the circuit, and the common mode signal rejection of the input will be reduced--thus making the cable much more vulnerable to interference.
I would suggest matching the resistors of each pad to within a very few Ohms of each other if you can. It doesn't usually matter much if the two pads in a pair don't have absolutely identical loss; a fraction of a dB one way or the other can even be used to your advantage, to reduce the inequality of sensitivity which nearly any two microphones will have--even a "matched pair." For that matter the record level controls of most preamps are only approximate in their calibration, so the "35 dB gain" setting may well be +/- 1 or 2 dB.
--best regards