Is County-Level Aggregation Sufficient To Anonymize? Is county-level aggregation sufficient to anonymize the data such as that found in the US Census or GSS?
 A: It is difficult to answer precisely without a more detailed information. However, in general aggregating might not be enough. Have a look to the concept of k-anonymity (also l-diversity and t-consistency). As David Smith says people with uncommon quasi-identifiers (combination of data such as gender,age,profession), can be easiliy disclosed.
A: No, not in general.
As a simple example, suppose you had access to the number of hospital visits broken out by income ranges released at the county level.
If the dataset revealed that 1 person from the $1,000,000+ income bracket visited the hospital, and only 1 person in the county had an income that was greater than that amount, then you have effectively revealed the identity of that patient. This problem is demonstrated in the table below using Loving County, TX (total population of 113) as an example.
+--------------------------+-------------------+
| County                   | Loving County, TX |
| Population $0k-1M Income | 112               |
| Visits $0k-1M Income     | 12                |
| Population $1M+ Income   | 1                 |
| Visits $1M+ Income       | 1                 |
+--------------------------+-------------------+

Specific knowledge of the data and statistical contemplation must be taken into account to ensure aggregated data is sufficiently anonymized.
