My understanding is that Gaussian is a slightly better program from the perspective of the computational chemist, but Spartan is easier for the causal user to use (a reason why WebMO has been popular-it's a user friendly "front end" to Gaussian). How much do you plan to use it for high end calcs vs. teaching?

I'm glad you're looking at including computational chem!   I think that Gaussian has a lot more features (methods, functionals, included basis sets) than Spartan, and is much better at running jobs which should be parallelized (big ones!) or jobs with Ln/Ac elements.  I guess the question is would you use the extra features?  Spartan is easier to use (lots of "click the box" features) than the command line version of Gaussian, but my sense is that (inorganic) students who go on to do compuations in grad school will probably use Gaussian.  I use Gaussian with WebMO as the "front end" graphical interface (so I don't need to teach my students UNIX) to get some of the ease of use features of Spartan with the more powerful (and useful for my research) Gaussian program.  Gaussian also sells an interface (GaussView) which should make it as easy to use as Spartan (my memory is that it doesn't come cheap--and WebMO did most of the same things, as well as interface with multiple computational programs, for less $$$).  My students (undergrad p-chem and inorganic) have had very little trouble learning to use it. 

And here's a great resource that Lori created for faculty to help with introducing computational chemistry to undergraduates!  This is one of our "5 slides about" learning objects entitled Basics of Computational Chemistry.  Buried in here are useful tips on Gaussian and WebMO!  If you use this or any of the other computational chemistry learning objects from VIPEr, please post a comment and let the community know how they worked for you.

You could also try looking into GAMESS - it is free of charge and is a very powerful program.

http://www.msg.chem.iastate.edu/gamess/

 It is very easy to build input files using free molecular modeling programs such as:

MacMolPlt (free of charge)

https://brettbode.github.io/wxmacmolplt/MacMolPlt_Manual.html

and Avogadro (free of charge)

http://avogadro.openmolecules.net/wiki/Main_Page