What exactly are you trying to measure? some of the great North American and Andean batholiths have cooling histories of at least 20 million years, and maybe more than 50 million years. It is perhaps inappropriate to think of the crystallization process as having a single date. Phenocrysts near the top of the pluton are cooler, but may sink to deeper levels where they are partially resorbed or mantled in multiple cycles, giving rise to zoned phenocrysts. It can be a very long process. Generally, late stage pegmatites, which have been enriched in volatiles that depress the melting temperature, will indicate the youngest age of crystallization. Deeper levels of the magma chamber will be older, but really one needs to study the specific field relations, including extent of metamorphic aureole to decide upon the most appropriate younger or older samples. The middle will most probably give a reasonable average date. Remember, what you are measuring is the time of that the mineral cools through its 'closure temperature', which will be slightly to significantly younger than its time of crystallization.
As to the method of isotopic dating, K-Ar is generally OK but the relative error bounds get worryingly large for dates < a million years. Ar-Ar is more precise but more expensive. It's also an excellent method, as is Rb-Sr, especially for really old precambrian plutons - provided the sample is fresh. Any weathering or diagenesis messes up the method.