Where I started for each and what I'd do differently
Details on how I'll present R code on this blog are here.
I grappled with this one a while and still do, so no good beginner book recommendations other than stay away from the pretty graphs in Access. I started with Microsoft Access and those point-and-click queries in ... 2006? Anyway, point-and-click got frustrating quickly, as that never did quite what I wanted. I was told to query properly. So I used JetSQL in Access. A few years ago on the advice of a real programmer, I migrated to postgreSQL and pgadmin.
What I'd do differently: start with postgreSQL instead of Access. It's more complicated, but so much easier to not have to convert your database from the partially automated drop-down menus in Access to an easier-to-query format later. postgreSQL also is more powerful, which I ran into recently while trying to help a colleague use JetSQL. Some things I was used to doing in postgreSQL were more complex or impossible in JetSQL. I'd also have read up a little on database structuring and have thought out my database structure more. I've had to do several big re-structurings of the architecture since I started. I think I've finally got it where I can easily integrate new tables if needed.
Long story short: it's easier and more efficient in the long run if you just go ahead and learn to query using SQL (Structured Query Language).
I learned using code snippets from a class in 2008 and, since then, deep ventures into the help files. I am trying to convert to R mostly, so mainly I will post SAS code to compare with R code. Using Multivariate Statistics (Tabachnik and Fidell 2007) has lots of good SAS code to use. So, sometimes it's simplest just to follow their examples if you have that book and a problem that it addresses.