Having been involved in admissions, I can attest that some people here are blowing a lot of smoke. GRE's are typically the least important component of an application.
GRE's are typically used for two purposes in admissions: 1) university-wide fellowship competition; 2) department admissions' minimum cut-off. Typically, anywhere from 1150-1200 on Q and V stands as an absolute minimum cut-off point for some, though not all, schools. Superb GRE scores (i.e., 1450+) can give you a boost, but they will never get you in in their own right. GRE scores are not used as the main basis for comparison of applicants. In fact, statement of purpose, letters of recommendation, and writing sample (where applicable) are the most important components of the applications and serve as the most reliable basis for comparison (someone could get very lucky on the GRE but evince no real talent for sociological study).
The confusion seems to be over the difference between average GRE scores of admitted and/or enrolled applicants and minimum required GRE scores. There is often, though not always, a correlation between successful applicants and applicants with high GRE scores, but this is certainly not a casual relation (as we all have learned in Sociology 101, correlations and causal relations are not coextensive).
Finally, let's look at UNC's published admissions stats in sociology (a top-5 program):
In 2009, the average GRE scores for ACCEPTED applicants were: 526.67 V, 633.33 Q. These are both well below the 90th percentile in their respective area. The average scores for DENIED applicants were: 587.14 V, 610.00 Q. Source: http://gradschool.unc.edu/pdf/2009-ADMISSION-STATISTICS.pdf
As you can see, UNC's denied applicants had a higher average GRE score in V than the applicants it accepted, and the Q scores were comparable. This is just one example (though there are many more), but it is enough to show that there is quite a bit of confusion in this thread on how graduate admissions work.