Kempthorne uses the randomization-distribution and the assumption of unit treatment additivity to produce a derived linear model , very similar to the textbook model discussed previously.  The test statistics of this derived linear model are closely approximated by the test statistics of an appropriate normal linear model, according to approximation theorems and simulation studies.  However, there are differences. For example, the randomization-based analysis results in a small but (strictly) negative correlation between the observations.   In the randomization-based analysis, there is no assumption of a normal distribution and certainly no assumption of independence . On the contrary, the observations are dependent !
2015Dec14: New ANNOVAR version is available. The major change is to enable multi-threaded ANNOVAR for gene, region and filter annotation. Use an argument such as -thread 6 to turn on multi-threading with 6 CPU cores. Special thanks to S/W R&D Center, Device Solutions at SAMSUNG for proposing the changes and demonstrating the performance using Samsung 830 SSD drives. I re-implemented their algorithm, and confirmed that >10 fold speed improvement for filter annotation for genomes can be achieved even in a Seagate HDD RAID-6 array.