Page 300 - Invited Paper Session (IPS) - Volume 1
P. 300
IPS153 John D.
to be dealt with. In the Irish PECADO project the SPD is called the Person
Activity Register (PAR).
Estimate population size using DSE methods The PECADO project
identifies a list B derived from another administrative data source to use in
the DSE setup with the SPD as list A. List B is comprised of all those applying
for or renewing their driving licence in the calendar year. We call this list the
Driver Licence Dataset (DLD). The DLD is excluded from the SPD.
The methodology underpinning this approach is documented by Zhang
ˆ
and Dunne (2018). The population size estimator = / where
̂
, are the respective sizes of list B, list A and the match between list
A and list B, is underpinned by the following 3 assumptions
1. No erroneous records: A closed population ensures no records from
outside the population are included but we also suppose there are no
duplicate records or incorrectly identified records in either list A or list B.
2. Matching assumption: There is no linkage error when matching
records between list A and list B.
3. Homogeneous capture with respect to list B: Every unit i in the
population U has an equal chance of being captured in list B.
An additional assumption of independent capture in List B, that is the event
of any person is captured in list B has no impact on the likelihood of any other
̂ ̂
person being captured in list B, provides the variance estimator () =
(( − )( − ))/( ).
3
The DSE methods described here (Zhang and Dunne, 2018) provide for a
broader application of DSE than the more traditional approach as described by
Wolter (1986). In particular, we can now consider a DSE model when list A is
compiled from administrative data sources where it can be difficult to justify that
the traditional assumptions hold.
This system produces stock population estimates on an annual basis on a
population concept similar to the annual resident population (Lanzieri, 2013).
The annual resident population concept used here is based on a calendar year
rather than a point in time. The demographic accounting framework is adjusted
slightly to reflect this. The accounting identity becomes: population resident in
year (t) is equal to the population in year (t-1) less outflows in year (t-1) plus
inflows in year (t). Reliable stock estimates will provide reliable estimates of net
flows, that is the difference between inflows and outflows, but won’t provide
estimates of gross population flows. We propose an extension to the DSE
methodology above to estimate gross population flows building on the stock
estimates.
2.3. Estimating Gross Flows: Inflows and Outflows
The PECADO project reuses the underlying data sources from two
consecutive years, 1 and 2, to estimate gross population flows. Gross
289 | I S I W S C 2 0 1 9