I agree with Nigel's statement from a statistical standpoint. But to follow up on it a little more, does anyone know how the "tolerance" for passive components is defined?
That is, is the "+/-5%" (etc.) related to a coefficient of variation (i.e., the standard deviation divided by the mean as a percentage), with the limit set at 2, 3 or some other number of standard deviations, or is it more like a manufacturing tolerance for which one would expect all (>99.999%) of the components to fall within?
Apropos his comment, "assuming the resistors are from different batches," I have noticed that within a single apparent lot (e.g., a section of cut tape), the values are pretty close to each other, but not randomly distributed with respect to the expected mean. That observation is just anecdotal, but does relate to how one would calculate the expected error for a combination of components from the same lot.
John