Academic journal article Population

Completing Life Histories with Imputed Exit Dates: A Method for Historical Data from Passive Registration Systems

Academic journal article Population

Completing Life Histories with Imputed Exit Dates: A Method for Historical Data from Passive Registration Systems

Article excerpt

(ProQuest: ... denotes formulae omitted.)

The number and variety of longitudinal databases available for both historical and contemporary research have been growing rapidly, and these databases raise complex problems due to missing information, loss of response, and censoring not found in cross-sectional data sources. In this paper we present a way to impute missing dates of migration in event histories compiled from "passive registration" systems, such as genealogies.(1) This is a common problem in historical demographic research, and similar problems may occur in contemporary data sources, especially those incorporating information from administrative records.

In a "passive registration" system, subjects are not under continuous surveillance. Their presence in the study area is only known when an event (e.g. birth, death, tax payment, hospital admission) is recorded. Databases of this kind can be constructed from administrative sources that report many kinds of events, like visiting a physician, applying for insurance, or using a service. If individuals can move out of the registration area without reporting the date of their departure, the ending dates (censoring times) of those life histories are unobserved. This problem sometimes occurs in demographic and clinical research, and it is common when life histories are taken from historical sources. The classic example is "family reconstitution", in which life histories are reconstructed by linking together births, marriages, and deaths drawn from church registers or vital registration. Since the geographic scope of the documents is usually limited, life histories are censored when individuals move to a different administrative area (e.g. parish or municipality). The date of censoring is rarely known, however, because dates of migration are almost always lacking.

Historical demographers are increasingly supplementing vital registers (births, deaths, and marriages) with other sources (like censuses, tax registers, and city directories) to obtain more information about time under observation. Gutmann and Fliess (1993) and Garrett and Davies (2003) describe datasets in which births, marriages, and deaths are linked to censuses. Individuals and families who appear in two consecutive censuses are at risk of events for the entire period, but some events are linked to families appearing in only one of the two censuses. The time at risk of families found in the first census but not in the second is not known exactly, but it must be less than the time between the last record in the vital registers and the second census.

We also encounter this problem in population registers, when information about migration is under-reported. A number of large databases derived from population registers are becoming available, such as the Demographic Data Base at Umeå University and the Historical Sample of the Netherlands.(2) In principle, population registers record movements between communities, but information on migration is sometimes missing. When Belgium established population registers in 1846, migrants were required by law to report changes of residence at both their origin and destination. It soon became clear, however, that people were less likely to report leaving a municipality than establishing residency in their new home. A previous study of nineteenth-century population registers from a Belgian city found that 20% of migrants disappeared without reporting their departures (Alter, 1988). One of the advantages of population registers over family reconstitution is that they record information about migrants as well as non-migrants.

This paper describes a new procedure for imputing migration dates in order to complete life histories. Our method combines elements from several earlier approaches, which are described in the next section. We also present tests of our procedure by applying it to data from nineteenth-century Belgian population registers in which dates of migration were reported. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.