sadly I think I tend to reach for those plug-ins I am most familiar with, and being a bit of a dinosaur, that means I tend to reach for emulations of the hardware I grew up with.

There are exceptions to the dinosaur rule, of course, but familiarity is important.

I know what Echoboy can do, I know how to use a Urei 1176 or dBX 160, and so on.

I do try to spend time with new plugins, especially if the might offer capabilities I don't already have. It's always an eye opener, and sometimes a cool surprise awaits!

The Cakewalk Transient Shaper, for example, does stuff the SPL version doesn't. I still use the SPL version most of the time, because I am familiar with it, but if I need to something more extreme then  I'll use the Cakewalk version.

Don't want to get into all these different preferences but to add on what WST3 stated, one must know how to use each tool and what makes one different than the other.  The only thing that can give you this knowledge is experience and some teaching from other sources (other engineers, the internet, books, other recordings, etc.).  But still, even with the teachings, experience is the only way to really understand your tools and what makes each one different.
 
Understanding SOUND is also important... one of the many things that took me awhile to get my head around was you really can't add something that isn't there in the first place (with EQ) but you can take out some of which is there.  So, how you record the track is as important as how you mix the track.  And the other important thing about compression is how the attack and release work together and then how each unit has different characteristics. For example, you can compress the AIR out of a track without slamming the input gain easier and more effective on some compressors versus others. I would say that the 1176, DBX, LA2a and LA3a, Fairchild and a few other vintage compressors are easier to use and as such easier to gain experience before starting with the more advanced modern compressors like the MANLEY and the VSC. 
 
Like most things, it is best to start with the basics and build upon your own knowledge.
 
IMHO
 
 

Experience is the major part of it.  Over time as you try different plug-ins, you get a feel for how they affect the sound, how intuitively the controls respond to you, which types of tracks they work best on, etc..  There are no right answers and ultimately it comes down to personal preferences.  
 
Take EQ for an example.  Some people love the visual aspect of EQs like QuadCurve where they can see the source frequency spectrum and the EQ curve and really micromanage the frequency response.  Others hate that level of detail and prefer stuff like the Blue Tubes EQs were you get some broader knobs for "Bass", "Treble", etc. and have to use your ears more to understand what the controls are doing.  The same distinction applies for compressors, delays, etc..  
 
I advise trying to understand both approaches and don't be afraid to stack them.  For example, for distorted guitar tracks, I find something like the Blue Tubes Pulse-Tec EQs to be great for dialing in the meat of the track -- it's like having an amp tone stack available post-recording -- but I'll typically also use something like the QuadCurve in addition for surgically removing annoying frequencies, noise, etc..   
 
Also, if you spend time on forums like this, reading books, watching instructional videos, etc., you'll pick up tips from others and their experiments.  Again, it's personal preference so a lot of it might not resonate with you but it is still nice for pointing you in the right direction ("Try plug-in X on Y style tracks!").

jimkleban
Understanding SOUND is also important... one of the many things that took me awhile to get my head around was you really can't add something that isn't there in the first place (with EQ) but you can take out some of which is there.

So true! Especially the part about you can't add what isn't there. You can subtract everything else to make it look like there is more energy in a specific band, but that's it! (Exciters can, in fact add, but that's a different topic<G>!)
 

jimklebanSo, how you record the track is as important as how you mix the track.  

That is the place to start! I was really fortunate that one of my mentors emphasized (over and over and over) the importance of microphone selection and placement. These two words were always used together, the microphone you selected affected the placement, or if you had limited place to position the microphone that would affect selection. Two sides of the same coin.

Which make sense, but the thing that made his recordings remarkable was that he used selection and placement as the first stages of equalization and dynamics control. He would not reach for a knob until he had the microphone exactly where it needed to be. That lesson still sticks with me on those occasions when I am fortunate enough to record live players!
 
The analog to that would be selection of synthesizers or sample libraries, and even specific patches. You don't have as much control (the developer made most of the decisions for you already), and it does get expensive, but that's what we have to work with when we turn to virtual instruments.
 

jimklebanAnd the other important thing about compression is how the attack and release work together and then how each unit has different characteristics.

 
Read the manuals, or look at the front panels!
 
The Urei 1176, for example, has a fixed threshold, you determine the relationship between your signal and the threshold by adjusting the input level. Pretty cool really! First, it means you can use that input stage to color the sound, second, when one designs a compressor there are all sorts of trade-offs to be made (e.g. feed forward or feed back, variable gain or attenuation, and so on.) Choosing a fixed threshold left Mr. Putnam free to address some of the artifacts from earlier designs with different solutions. So the 1176 behaves quite differently than the dBX 160 because Mr. Putnam and Mr. Blackmer had very different approaches to the problem.

When I use an 1176 I first "calibrate" it - habit I suppose, but I set it up for unity gain with some small degree of gain reduction, then I know what's going to happen when I spin various knobs or push buttons. I can, for example, increase the input level and decrease the output level to maintain unity gain through the device, but by doing so I'm also changing where  the input signal crosses the threshold.

It isn't as easy to do that with the dBX 160, so my approach to that is to set the threshold to the maximum and set levels accordingly, then start to apply gain reduction by way of the variable threshold.
 
That's where I suggest folks start, set the attack and release fast so the level changes are really audible and play with the threshold. Then start to play with attack and release times. Which is an entire topic in and of itself!
 
Last thought for now - equalizers also have  quirks! The Valley People Maxi-Q placed the filters in parallel, while most other parametric equalizers place them in series. So they operate differently, believe it or not! One of my favorites, the Maag EQ4, has a band that exceeds the audible spectrum by a couple of octaves, it tends to sound really smooth. There are lots of little tricks like that!

Time spent with an experienced engineer is still the best teacher, but there are some reputable courses out there. When asked I recommend Dave Moulton's Golden Ears.