I do not think ADP goes back and adjusts the numbers from its previous reports. I think this is because ADP does not use sampling - ADP uses the actual data from the private companies for which it provides payroll services. With this data is pretty clear, among many other things, how many people were employed at this particular company in this particular geography at this particular time. ADP then aggregates all of this data in its report.

Right -- I guess I'm wondering if ADP figures were lower than BLS figures by a proportionally significant margin (enough to possibly confirm these new values). I haven't been able to find the figures myself.