Summary of data linkage steps used to assemble the LongSHOT cohort
| Step | Blocking key | Focus of linkage algorithms and manual reviews | Step rationale | Percentage of all matches identified | |
| Purchasers-voter file (n≈1.2 m) | Deaths-voter file (n≈1.7 m) | ||||
| A | Same date of birth+same residential address | Name fields | Identifies highest probability matches | 80.12% | 79.74% |
| B | Same date of birth+similar first and last names | Name fields, address fields | Removes address as a blocking criterion to capture relocators between SVRD extract date and purchase/death date. | 17.92% | 15.64% |
| C | Same date of birth+both persons female+similar first and middle names | Name fields, address fields | Same as step B, except also allows changes of last names among women. | 0.73% | 0.06% |
| D | Same address | Name fields, date of birth field | Removes date of birth from blocking key to allow for errors in this field. | 1.24% | 4.56% |
LongSHOT, Longitudinal Study of Handgun Ownership and Transfer; SVRD, Statewide Voter Registration Database.