In a steel frame airplane or really any airplane you will want to consider paying the extra for a compass like Sirs that has corrections for coefficients A, B C, and D.
Coefficient:
A: will be a consistent error, and you can just rotate the indicator
B: will fix the front to back magnetism, when you swing your plane on a compass rose it will be an error that is the sine of your local flux lines.
C: Is like B but is caused by athwartship magnetism and will produce an error that is the cosine of the heading.
D: Is typically caused by electronics and will typically be twice the sine of the heading.
E: Is caused by soft iron and will be twice the cosine and shouldn't be an large issue on tube and frame aircraft and I don't think any common aircraft compasses offer this correction anyway but is usually fixed with large iron spheres on ships.
Take your plane to an airport with a good compass calibration pad; turn on all of your gear and have someone help you swing your plane and chart the error you can get way more accurate than most cards will be set even in commercial planes.
You will need to find someone who is good at math to help you out, but the error will approximately model the following. One of these days someone should produce a phone app or web page that will let you play with Fourier transforms to guess the corrections, but by hand rotating the plane with a plumb bobs taped to the nose and tail to line up on the lines you can work them out through brute force.
Deviation ≈ A + B sin θ + C cos θ + D sin 2*θ + E cos 2*θ
If you use a compass with just N-S and E-W adjusters (like most airpath models) you will need to depend on the card. If you spend the extra money on a model with 4 magnet correction abilities or better it will take a bit more effort to set but will be far nicer in the long run (sirs as an example).
I think the cost difference is around $100 so weigh that with the convenience of not needing a card.