When I started this thread, I wasn't really sold on national unity as the topic of the thread. Mechanically it doesn't really make sense because it covers too many factors.
There are many facets of war fighting. Is it an offensive war? Is it a defensive war? The United States could certainly lose enough troops in Europe to sign peace with whoever they were fighting. However, in the face of an invasion the USA would be an entirely different animal. You could see 10-20 million losses and the country is going to keep fighting.
A dissent mechanic should probably be introduced, and national unity if it isn't already tracked on a per-conflict basis should be.
Dissent would function as a penalty to industry/logistics etc.
So the way I see this working is, casualties impact national unity and dissent. Dissent is universal, national unity is per conflict. Casualties taken on home soil cause zero, or next to no national unity and some degree of dissent. Casualties taken on foreign soil (neighbor) cause some dissent and some national unity penalties to THAT specific conflict. Casualties suffered overseas cause significant dissent and significant national unity penalties to that specific conflict.
In the case of a massive world war, the USA losing a million men in India is going to be a lot different than the USSR losing a million men in Moscow. The repercussions for both of these situations are radically different. Likewise, the USSR losing a million men invading Japan, versus the USA losing a million men defending Boston. Again, the USSR would have internal problems with such losses, whereas the USA would have virtually no problems internally as a result of defending Boston.
Whether or not these suggestions are realistic additions, I think the depth of game-play would be expanded greatly by looking at some way to involved these ideas in the strategic layer. It's just wrong for the USA to lose 50 divisions or so in failed invasions and go about business as usual.