Volume 28 - Article 23 | Pages 649-680
Chronological objects in demographic research
|Date received:||28 Feb 2012|
|Date published:||27 Mar 2013|
|Keywords:||age measurement, calendar, chronological, dates, duration|
Background: Calendar time, age and duration are chronological objects. They represent an instant or a time period. Age and duration are usually expressed in units with varying lengths. The number of days in a month or a year depends on the position on the calendar. The units are also not homogeneous and the structure influences measurement. One solution, common in demography, is to use units that are large enough for the results not to be seriously affected by differences in length and structure. Another approach is to take the idiosyncrasy of calendars into account and to work directly with calendar dates. The technology that enables logical and arithmetic operations on dates is available.
Objective: To illustrate logical and arithmetic operations on dates and conversions between time measurements.
Methods: Software packages include utilities to process dates. I use existing and a few new utilities in R to illustrate operations on dates and conversions between calendar dates and elapsed time since a reference moment or a reference event. Three demographic applications are presented. The first is the impact of preferences for dates and days on demographic indicators. The second is event history analysis with time-varying covariates. The third is microsimulation of life histories in continuous time.
Conclusions: The technology exists to perform operations directly on dates, enabling more precise calculations of duration and elapsed time in demographic analysis. It eliminates the need for (a) approximations and (b) transformations of dates, such as Century Month Code, that are convenient for computing durations but are a barrier to interpretation. Operations on dates, such as the computation of age, should consider time units of varying length.
Frans Willekens - Nederlands Interdisciplinair Demografisch Instituut (NIDI), Netherlands
Other articles by the same author/authors in Demographic Research