How do you use statistical sampling to argue that restoring backup tapes or reviewing a data set is not proportional?

Statistical sampling lets a party support a proportionality argument with evidence instead of mere assertion. Rather than claiming that restoring backup tapes or reviewing a data set “would cost too much,” you process a small, representative subset and use the measured results to project the full burden and likely value. US federal civil discovery weighs proportionality factors such as the amount in controversy, the importance of the issues, the parties’ resources, and whether the burden outweighs the likely benefit. Sampling produces the numbers that feed that balancing test. Rules and standards vary by jurisdiction, so confirm the framework that governs your matter.

How sampling builds the argument

A defensible sample typically follows a few steps:

• Define the population. Identify the tapes, custodians, or repositories at issue and their approximate volume.
• Draw a representative sample. Use random or statistically valid selection so results can be projected to the whole.
• Restore and review the sample. Capture the actual time, cost, and technical effort (some legacy backups require specialized restoration).
• Measure responsiveness. Record how many sampled items are relevant, unique, and not already available from more accessible sources.

Turning results into a proportionality showing

From the sample you can extrapolate two things courts care about:

1. Burden and cost. Multiply per-unit restoration and review costs across the full population to project total expense and time.
2. Marginal value. If the sample yields few responsive, non-duplicative documents, you can argue the remaining data is unlikely to produce meaningful new evidence.

A low hit rate paired with a high projected cost is a strong basis to argue the effort is disproportionate, especially where the same information exists in active systems. Sampling is also central to “not reasonably accessible” arguments for sources like disaster-recovery tapes.

Practical cautions

Document your methodology so it is repeatable and transparent. Disclose the sampling approach to the opposing party and, where appropriate, the court early; cooperative, defensible methods carry far more weight than unilateral conclusions. Be prepared to adjust if results show the data is more valuable than expected. For related concepts, see our overview of e-discovery.