Some more words on capacitance and discharge lamp flicker. I have a drawing at home, and if you email me (see profile) I can send you a copy.
This is a relaxation oscillation, similar to the old neon timebase, (search google for neon timebase gets you this ..http://www.vego.nl/4/n/n_029.htm) but basically the voltage rises when the lamp is off, as the impedance is many megohms. Once threshold is reached the gas breaks down and the lamp lights, but the source impedance is far too high to sustain action for more than a brief flicker, so the volts on the capacitor plummet and the light goes out, and back to high impedance state. Capacitor starts to fill again....
Now leaping forward from 1931 to the present day, and AC mains, the high impedance supply is the capicitance of the wiring, 100p/metre for 1mm T+E, so ~30 jMeg ohms for every metre, so say 10m, say 3meg.. say 100uA of charging current, and the old neon is now replaced by a rectifier and oscillator circuit of the compact flourescent lamp (CFL). The reservoir capacitance downstream of the bridge in the lamp will need to reach say 50V to start, and will be perhaps 100uF, so flicker rate of half a second, say. result irritating flicker, worse when tube is warm as strike voltage is lower.
If you set this up on the bench with nothing other than an open ended length of cable (as if to go to a switch , but don't fit one) and the CFL, by varying the length of the cable, and therfore the charging current, you can alter the flicker rate.
If it is 10m or so of wire in the switch circuit, then a permanent fix is either to place low current resistive load (perhaps 100K or so - best made up of two resistors in series, to prevent a single failure), or a nanofarad or 10 of suitibly mains rated capacitor in parallel with affected device, to pull the voltage accross the lamp firmly down when off. If you prefer ready made componenents you might find a switch suppressor accross the lamp (not the switch -that would increase the leakage) is all you need. I'd suggest building these into the lumiaire if there is room, or a proper joint box, as dangling live components in the breeze will attract the wrath of the rightous (though its OK inside washing machines and TVs..).
I suppose if you were feeling cheapskate then the parallell capacitance could just be an extra long cable feeding the lamp.
Rest assured this is a well known problem, and need not indicate wiring fault. (though in switched neutral, rather than switched live, the effect can be much worse, as now the total capacitance to earth adds to rather than reduces the leakage current.)
If this makes no sense, come back to me
Mike P_J