No, that's not what I said, and I apologize for not being clearer.
To use your original example, say you want to layer Voice 001 with Voice 002.
This way should work:
(1) Select Voice 001 as the Main Voice.
(2) Press DUAL to turn on the Dual Voice.
(3) Hold down DUAL to go to the Dual Voice function.
(4) Select Voice 002 as the Dual Voice.
This way will most likely not work:
(1) Press DUAL to turn on the Dual Voice.
(2) Hold down DUAL to go to the Dual Voice function.
(3) Select Voice 002 as the Dual Voice.
(4) Select Voice 001 as the Main Voice.
If you do it the first way, selecting Voice 001 as the Main Voice will most likely turn off the Dual Voice and (even though Dual Voice was just turned off) set the Dual Voice to some voice that has been preselected as going well with Voice 001-- which, going by your first post, is apparently Voice 054. Once the keyboard loads the presets for Voice 001, you're free to override them by changing the Main Voice settings if desired (Volume, Octave, Reverb Depth, etc.), turning on the Dual Voice if desired, changing the Dual Voice to Voice 002 if desired, and changing the Dual Voice settings if desired (Volume, Octave, Reverb Depth, etc.).
But if you do it the second way, everything you did in steps 1 through 3 will be undone by step 4, because when you select Voice 001 as the Main Voice, the keyboard will automatically turn off the Dual Voice (even though you had specifically turned it on in step 1), and automatically set the Dual Voice to Voice 054 (even though you had specifically set it to Voice 002 in step 3).
However, this behavior varies slightly from model to model, depending on the types of preset voices which are available on the keyboard.
For instance, on my PSR-E433, PSR-E443, and PSR-EW400, it works just as I've described above, because those models include several "dual" voices in their presets.
But my YPT-400 (which is another name for the PSR-E403) does not include any "dual" voices in its presets, and selecting the Main Voice does not automatically change the Dual On/Off and Dual Voice settings.
Looking at the Voice List for the PSR-E363/PSR-EW300/YPT-360, I see that these models include "dual" voices (numbers 162 through 188) and "arpeggio" voices (numbers 216 through 235). Therefore these models should work as I've described above with respect to the Dual and Harmony/Arpeggio features-- i.e., whenever you select a Main Voice, the keyboard will automatically change the Dual On/Off setting (on for voices 162 through 188 and off for the rest), the Dual Voice setting (preselecting a voice that goes well with the Main Voice you selected), the Harmony/Arpeggio On/Off setting (on for voices 216 through 235 and off for the rest), and the Harmony/Arpeggio Type setting. Other settings related to those functions will also be affected-- Main Voice Volume, Main Voice Octave, Main Voice Chorus Depth, Dual Voice Volume, Dual Voice Octave, Dual Voice Chorus Depth, Harmony Volume, Arpeggio Velocity, and Pedal Function. You can change those functions to any settings you want, but you should always do so AFTER you've selected the Main Voice you want to use.
By the way, choosing a Style or a Song will also affect certain settings-- in particular, the Tempo and the Time Signature (Time Signature Numerator and Time Signature Denominator). Once you've selected a particular Style or Song, you can change the Tempo setting to speed up or slow down the Style or Song playback as desired, but changing to a different Style or Song will set the Tempo to the default for the new Style or Song. And the Time Signature is locked for each Style and Song, so you won't be able to change it at all. However, if you want to play or record a Song of your own then you will be able to set the Time Signature as desired-- unless you're using a Style in your recording, in which case the Style's Time Signature will be used.
Also, if you set the Main Voice to Voice 000, the keyboard will automatically change the Main Voice to one that's been preselected for the selected Style or Song.