You may be misunderstanding what SK was saying. Look at this:
http://en.wikipedia.org/wiki/Magnetic_Declination#Change_of_declination_over_time_and_space
Looking at the illustrations, it should be obvious to you that there is no 1-to-1 mapping between magnetic coordinates and geographical ones. That is two people thousands of miles apart may measure the same magnetic latitude. It isn't a good predictor of where you are, in general. It is best used locally.
I am not motivated to actually do the algebra to give you the answer, but this tells you how to get the answer.
Any point on the surface of a sphere can be represented by two coordinates which are transformed in a two by two matrix into any other (arbitrary) pair of coordinates. In 3 space, R is constant (approx) so what I would do to solve this problem is convert figure out the two matrices which convert geographical coordinates to spherical φ and θ, and do the same with your magnetic coordinates. Then figure out the inverses and multiply the Inverse of one by the other and you'll have the two transformation matices.
In other words given (lat, long) there is a 2x2 matrix which will convert them to (θ,φ). And there is a matrix which will convert (θ,φ) to (mag_lat,long) ((hint the final matrix will have for long (assuming geograph and geomagnet are the same) the row 0...1)). So, all you need figure is the values of lat and long which will transform to mag_lat. It will be composed of sin() and cos() functions of the (θ,φ) coordinates.