It's as if Page read my mind on this one.
I'll parrot his conclusion that it's more work, but can be rewarding in the challenge/education/enjoyment of the effort to do it, and possibly in the sound of the finished result too.
I'll add two things-
First, do as Page suggests: listen, EQ separately, decide, relisten, play with mix levels, and listen again. That is usually enough in itself. If you want to try a further step (which I've thought through and consider doing far more often than I usually actually get around to trying) you can setup reciprocal high and low-pass filters on each- somewhat analogous to crossover filtering between speaker divers- so that there is less interaction between the two sources outside the crossover region. The easiest way to do so is probably to insert a new parametric EQ plugin instance on each source and set a single, identical corner frequency point in both. Change the filtering from typical peak or shelf filtering to high-pass for one source and low-pass for the other and set the same slope for each if that's an option. Then shift that corner frequency up and down, being careful to set both filters to the same frequency. Find a frequency which works best and then go on to play with the filter slopes if available. Once you find something you like, switch those EQ filters in and out (together) to compare the cross-over version with the simply summed version. Even if you prefer the simply summed version, you will have learned a lot while doing it. (at least I do!)
Some people do something similar to this to specifically compensate for the limited response of one source. For instance, WifiJeff here at TS often records with hypercardioid DPA 4099's (actually an earlier similar model DPA miniature hypercardioid mic) which have a typical rolled off low frequency response when used at a distance, co-located with 4060 omnis (actually 4063 if I recall) which have excellent flat bass response to the lowest octave but lack the directionality of the hypers. He mixes the resulting channels together to get 'the best of both' and uses EQ filtering to eliminate any interaction interference between the two outside of the crossover region. In that case I believe he doesn't high pass the hypercardioids since they are already naturally 'high-passed', but low passes the omnis with a slope that approximates the inverse of the hypercard response. The result is focused high frequency directional pickup from the hypers with an extended low frequency response that transitions to omni below several hundred hertz.