I do not consider any possibility that the Current Population Survey conducted by the U.S. Census Bureau for September 2012 is biased by CB or by the BLS, This is not the case. There is another hypothetical way to bias the data. There are around 70000 households surveyed by the CB. These households include approximately 200000 persons (mean household is 2.5 people). All these people (excluding several percent not responding ones) answer a few questions associated with their current status: employed, unemployed or not in the labor force. There current level of civilian labor force is approximately 155,000,00 with 12,000,000 unemployed. These figures are calculated by a projection of 200,000 to 310,000,000 using population controls. In essence, one person represents 1550 people.
How many people are needed to increase the rate of unemployment by 0.1%? The rate of unemployment is calculated as the ratio of the number of unemployed and labor force. So, 0.1% of unemployment rate with the level of labor force of 155,000,000 corresponds to 155000. Since one person in the CPS represents 1550 people, one needs only 100 people to increase the rate of unemployment by 0.1%. To decrease the rate by 0.3% , only 300 are needed.
I do not say that the result for September 2012 is biased. I say that the Current Population Survey procedure is wide-open for manipulations.