A family of Microsoft relational database management systems designed for ease of use.
To my mind the first step is to forget about MS Access per se, and to learn how, whatever relational database product is being used, the part of the real world with which a database is concerned is modelled in accordance with the formal principles of the database relational model. One of these is the process of normalization, which has already been mentioned, and this is certainly something which you will need to have a good grasp of, to a certain level at least. Normalization is a process which, when applied to a table ensures that the table does not contain redundancies which can lead to update anomalies. A simple example would be that a table of addresses should not contain columns for both city and state (or its equivalent), because it would then be possible for two or more rows in the table to include the same city, but erroneously put it in different states in each. Such a table should would be said to exhibit a transitive functional dependency on the primary key of the table, i.e. once we know the value of the key we know the city, but this should then enable us to know the state, which is transitively determined by the key via city. We now need to prevent the wrong states being entered for the city in question. This is done by decomposing the table so that the addresses table tells us which city the address is in, but not the state. We are told which state the city is in in a separate cities table, which is related to the addresses table on CityID columns. We can't use city names because they can legitimately be duplicated, so we give each city a numeric CityID value as its primary key, and the addresses table has a CityID column as a foreign key. Consequently the transitive functional dependency on the key of addresses is eliminated, and the table is normalized to Third Normal Form (3NF).
In the above we have touched upon the other basic mechanism in designing a database, which is how the tables are related to each other. We saw that an Addresses table is related to a Cities table on CityID. As each state can contain many cities, but each city can be in only one state, the relationship type is one-to-many.
In other circumstances a relationship type might be one-to-one, e.g. an Employees table might be related to a SalesPersons table on EmployeeID. In this case the EmployeeID column in SalesPersons is both its primary key and a foreign key. This type of model is known as a type hierarchy.
A relationship type might be many-to-many, e.g. An Orders table is in a many-to-many relationship type with a Products table, because each order can be for may products, and each product might be in many orders. In this case the relationship type is modelled by creating a third table which resolves the relationship type into two one-to-many relationship types by having ProductID and OrderID foreign keys. It will also have other columns for the unit price of the product, the quantity ordered etc. A many-to-many relationship type between two tables like this is a binary relationship type, but many-to-many table can also be ternary, quaternary etc.
For examples of such relationship types You might like to take a look at DatabaseBasics.zip in my public databases folder at:
https://onedrive.live.com/?cid=44CC60D7FEA42912&id=44CC60D7FEA42912!169
Once you are comfortable with how the different relationship types are used, take a look at Relationships.zip in the same OneDrive folder. This takes things further by showing how relationship types are built up across a database to achieve the overall model.
You'll also find Normalization.zip in the same folder. This gives as simple as possible, but no more so, explanations of the various Normal Forms. Concentrate in gaining a good understanding of the first three Normal Forms to start with. In most case a table normalized to 3NF will also be normalized to the higher normal forms, which is not to say that you should ignore the higher forms, just leave them until you are completely comfortable with the first three.