I dont think it really matters all that much since most kits are developed to use either plasma or serum or both. It is always better to consider which samples the researchers in the literature used if you want to compare your results to someone elses in particular. I have used both serum and plasma to detect antioxidants in rats with great success.
it depends on the test because if your test reagent react with clotting factor you have to go for serum.IF serum and plasma value are different then you have to stick on normal values given in serum means serum.If ene difference os not there to minimize the cost of anticoagulant you can use serum