Matrix Stacks

The 4x4 matrix you use for the object IS the scale, translation, shear, etc...
If you want to rotate your object you would have to modify the current matrix.
You don't need to store multiple matrices to do this, you can have general methods which operate on a current matrix.

I would like more specific information like what tutorials you were looking at and a clearer explanation on what you are trying to do.

I really suggest taking a look at this:
http://www.gamedev.net/reference/programming/features/scenegraph/

Im just confused on how to render a complex model (human for example) using these matrix stacks

Matrices are a mathematical representation of an object's state.
A matrix is used to define the object in some coordinate system, that which is defined by the programmer.

Under OpenGL/Direct3D this is how it works.

Example:
An object located at a point called p(2, 3, 1, 1), the last value is needed because of homogenous coordinates.
Could be stored in a row order matrix like:

[0, 0, 0, 2]
[0, 0, 0, 3]
[0, 0, 0, 1]
[0, 0, 0, 1]

Now, say that there is an object that is relative to this point, like a human arm relative to the hand.

So relative to the above point is another point in the world: P2(3, 3, 1).
This point is translated 1 unit in the x direction relative to the above point.

The matrix to represent this point is:

[0, 0, 0, 3]
[0, 0, 0, 3]
[0, 0, 0, 1]
[0, 0, 0, 1]

Multiply these matrices in the order I've mentioned to get where P2 is in the world.
However these matrices don't contain rotations or scale.

Most often homogeneous matrices are used to describe transformations. Those transformations are concatenated by multiplying the matrices. Hence a multiplicative identity is needed. The identity transformation is hence the identity matrix:

1 0 0 00 1 0 00 0 1 00 0 0 1

(K_I_L_E_R's matrices aren't correct in this sense.)