Sound Check values are calculated based on stored ReplayGain values. The Sound Check offset setting can be used to change the parameters used to produce the value. The offset, if any is relative to the LUFS value used to produce the ReplayGain Track Gain and ReplayGain Track Peak items. While any reasonable LUFS value can be used as a baseline for loudness, it makes more sense if all your tracks are based on the same value. As mentioned above, Yate's default LUFS value used in the Replay Gain calculations is -18. iTunes supposedly uses -16 as its LUFS value when doing its calculations. You can leave the replay gain LUFS as -18 and specify a Sound Check offset of 2 if you wish the Sound Check settings to be based on -16.

Typically I don't mess with these values. However if your LUFS value is -14 I would think that the Sound Check offset would be -2.

The sound check values are stored as a UDTI named iTunNORM for MPEG-4 files and as Comment (iTunNORM) for other supported files. These is where Apple expects them. Hint: you can find this info via Help>Quick Reference>Tag Mapping Table

Any downside changing these values from -18 LUFS?

As I said, I don't mess with them. However, as the audio stream is not modified, you can always simply recalculate the values if you're not happy with the results.

As part of my workflow I calculate the ReplayGain settings at their default values.

I always use the album Today by Mississippi John Hurt as a test when I play with the values. Regardless as to what settings I choose, the album always plays much quieter than anything around it. Personally, I find ReplayGain makes very little difference. Just my opinion.