In addition to the above prequations to eliminate spurious measurements you should also check that any capacitance due to barrier formation at the contact is not present. For this purpose, you should use different samples of the same material of varying thickness.The measured capacitance should be inversely proportional to thickness that is dielectric constant should not depend on thickness. If it is not so, you have to improve your contact.
The simple and for most purposes sufficient compensation procedure is to measure (in 4 terminal pair configuration - see Oskar's note above) "Short", "Open" and "Load" impedances that will enable you to determine all the stray impedances in the system ("T" topology). In Agilent equipment, this is done "automatically" by following "compensation procedure". Otherwise you will have to write down the equations for the three unknown stray impedances in terms of the three above mentioned measurements. This is more or less Agilent compensation procedure.
In order to measure true bulk dielectric "constant" of your polymer samples, you will have to determine in what frequency region you are measuring electrode-sample interfaces and in which frequency region you are measuring bulk properties. As pointed out by A.Kumar above, increasing the thickness of the sample by a factor 2 will increase the resistance (real part of complex impedance) of the bulk by a factor 2 , while decreasing the bulk capacitance (real part of complex capacitance) by factor of 2. You do not have to worry then about the ohmicity of your contacts.
To avoid the effect of moisture if any and also to shi eld your sample from electromagnetic waves, one should mount the sample in metallic sample holder with a good vacuum inside. Not following these prequations may lead to spurious dielectric constant and dielectric loss measurements.