S3U

History / Historical sources / Graduate / hse01

Archival Survival and Partial Identification

Bound a historical claim when record survival is selective.

All worksheets
5 questions0m 0s

Prerequisites

Source criticism, sampling versus selection, proportions, and bounds.

Learn the skill

Distinguish a surviving archive from the population of records originally produced. If survival depends on content, observed proportions need not represent the original collection. Bounds can express what is identified without inventing a missing-data mechanism.

Worked example

If 30 of 50 surviving petitions support a policy, 60% describes survivors. Without knowing which records were lost, it does not establish support among all original petitions, still less among all residents.

Model and Assumptions

Fictional archive: an inventory establishes 100 original petitions. Forty survive, of which 30 explicitly support a proposal and 10 oppose it. Sixty are unavailable. Assume each original petition was either supportive or opposed, with no duplicates; survival may be selective.

1. What percentage of surviving petitions is supportive?
2. What is the smallest possible supportive percentage among all 100 petitions?
3. What is the largest possible supportive percentage among all 100 petitions?
4. What additional assumption would justify using the surviving fraction as an estimator of the original fraction?
5. Do bounds for petition contents identify the share of residents supporting the proposal?

Further inquiry

Derive bounds when the number of original petitions is known only to lie between 90 and 110. Then specify which evidence could narrow survival uncertainty without assuming that silence means opposition.

Review criteria

  • Keep the 40 observed petitions fixed while varying the total.
  • Derive a lower bound of 30/110 and an upper bound of 100/110.
  • Separate record survival, record production, and population opinion.