I don't have exact numbers, but there is a correlation between the two- it just isn't the only one.  Your overall technological development also plays a part (it's displayed during the fight for that reason), as does the invasion tactic (which only acts as a modifier, but it's a huge modifier).

If your soldiering is less than that of the enemy, then an invasion tactic that acts against them, such as Tidal Disruption, can help by dramatically reducing their range of values for War Advantage.  Likewise, if you've got the edge, you can really crank up your War Advantage with Mass Drivers.

Once you've got this much down, it becomes fairly easy to guess how the fight will turn out before you invade.  Low soldiering can be overcome with good invasion tactics, larger troop numbers, or higher overall technology, roughly in that order of effectiveness.

Bonus tip: You can combine multiple transports in a fleet and attack with all of them at once to get a much bigger punch.

Bonus bonus tip: It's sometimes more effective to send in one transport with Mass Drivers and have them lose, so the planet becomes depopulated enough for a conventional tactic to easily take it.  Lower overall losses and a better planet, what more could you want?

I make it a priority to get the Triquan(?) Training wonder done... I also take a soldiering bonus in the startup for my custom race. In addition, I always take tidal disruption on the attack too, not just for the bonus but also to 'hopefully' destroy a buch of those race specific buildings I don't want. The only thig better is mini soldiers if you have the right tech tree... I routinely attack outnumbered 4:1 or 5:1, but have a 5-10:1 soldier number bonus. I had a 43 to 3 number roll the other day, 1,500 vs. 14,000. I had 842 men left after taking the planet...

I don't think I have ever paid much attention to the other numbers...

T

I used to get things like charging stalks and such left behind... perhaps this has been fixed and I just haven't been paying attention???

Not to mention the the odd manner the AI uses to place buildings anyways. Saves me the few seconds of dumbfounded shaking of the head when you see how pathetically some of these AI planets are done.  

That and I just like a clean slate so I can build it up the way I like it. So much easier with 10 empty tiles. I imagine I may have inadvertantly lost something important, but usually I check to be sure I'm not blowing a tech capitol or somesuch away before I go in... I imagine in a world where I took this some kind of serious I would be fanatical about trying to keep everything I could and all that, but for now I just like the silence of no buildings and no alien people about.

T

I *think* one-per-civ buildings are immune to being destroyed like that. Certainly Galactic Wonders are.

I usually try to keep as many of the AI buildings as I can, simply for the step up in building my stuff. No, I have no intention of keeping your slave pits, or charging stalks, or whatever else you have. I do intend to get whatever production they put out while building real factories and stock markets, and building as an upgrade is usually cheaper and faster than starting from bare ground.

You get credit for how much of the original building was built. It takes fewer MP to build an Industrial Sector on a Manufacturing Center than it does to build it on bare ground next to the existing MC (assuming the MC was built, not bought with no MP put into it). This means fewer turns to build the IS (faster) and fewer MP used to build it (cheaper).

Wow, i didn't know that.  So what you're saying is skipping entire buildings is absolutely best.  IE, don't stop at banks, or invention center just push on.

I had always thought it was best to push on, but I didn't realize how much better.

That almost seems like a bug if it's true.

Not true, and easily proven so. Hell, the game I was just playing had numerous instances of this. Assuming the industrial sector is completely constructed, upgrading to a CIC costs 6 bc, or one turn of construction (assuming at least 1 MP). Building the CIC from scratch might take 30+ turns depending on MP available, or cost ~3200 bc to buy.

I didn't test every single improvement over every other improvement.

In particular, it seems I tested primarily (if not entirely) early game improvements.

Which would explain it, actually.

All the early game improvements cost 50 production or less, so I wasn't seeing any significant difference in production between them.  The upgrade cost is assumed to be 50 instead (in most of the examples I ran), therefore you don't actually save any money, time, or production by building over them with an "equivalent".

And I just now noticed this by attempting to build an embassy over a traditional factory (which would cost me 50) versus building an embassy on bare ground (which would cost me 30).

So you have new building cost minus initial building cost (actually if rush buying is any indicator it uses the total social production spent on it, not the cost), but if the new building is actually cheaper than the initial building, then you pay 50 production to upgrade it.

I apologize for not being more thorough.

So does the starport.  It's because the "upgrade" costs less than the thing you're upgrading over, so it defaults to an upgrade cost of 50 production.

Also, for the record, it's 438 BC.