• For each dimension in your tabulation, summarize the data across the other dimensions. For example, if you have a 5 dimensional tabulation, you will now have 5 one-dimensional tabulations, in addition to your 5 dimensional tabulation.
• Find a reliable source of one-dimensional tabulations, and match the one-dimensional tabulations against the tabulations from the reliable source. Obviously, the tabulations must all use the same data source, and any recodes, filters, and geographical restrictions must be accounted for.
• If all of the corresponding one-dimensional tabulations match, then the multidimensional tabulation (from which the one-dimensional tabulations were created) must be correct.

If you need to verify a simulation:

• Request a tabulation using the same dimensions as your simulation (this tabulation is for verification - it is not necessarily the way you want to view the data). Include any volumes that you based the simulation on, and also the result of the simulation.
• Load the tabulation into a spreadsheet, and calculate the percentage differences between the simulation and the basis for the simulation. The differences should equal the simulation factors that you supplied.

There are an infinite number of user-defined dimensions that can be created, and we cannot create a set of rules for verifying all of them. As an example, we'll verify a dimension showing the percentage of Hispanic occupants of a household.

• Verify all other dimensions as described above, for simple tabulations.
• Request a 3-dimensional tabulation, where 1 dimension is the number of Hispanic occupants, 1 dimension is the total number of occupants, and the third dimension is the percentage of Hispanic occupants. The third dimension is constructed exactly the same way as the user-defined dimension in question.
• The first two dimensions of this tabulation are simple dimensions, and should be verifiable against other sources.
• Put the 3-dimensional tabulation into a spreadsheet, and calculate (based on the first two dimensions) the percentage of Hispanic occupants in each household. The result of the calculation should match the third dimension.
• The third dimension is now verified by other sources, and it should match the user-defined dimension in the original tabulation, thus verifying the original tabulation.