An index-based approach to determine partnership in a register-based census

In the preparation of its first register-based census in 2021, Estonia faces a challenge of forming dwelling-based households with inaccurate place of residence data. A pilot census in 2016 overestimated the number of lone parents by 67%. To increase accuracy, an index to measure probability of part...

Full description

Saved in:
Bibliographic Details
Published inStatistical journal of the IAOS Vol. 35; no. 2; pp. 245 - 251
Main Author Visk, Helle
Format Journal Article
LanguageEnglish
Published Amsterdam IOS Press BV 2019
Subjects
Accuracy
Algorithms
Census
Censuses
Children
Graph matching
Graph theory
Households
Model accuracy
Parameter estimation
Parents
Regression models
Statistical analysis
Online AccessGet full text

Cover

More Information
Summary:In the preparation of its first register-based census in 2021, Estonia faces a challenge of forming dwelling-based households with inaccurate place of residence data. A pilot census in 2016 overestimated the number of lone parents by 67%. To increase accuracy, an index to measure probability of partnership is being developed. Data on signs of partnership have been collected (marriages, divorces, mutual children, co-ownership of property etc.). In this paper, the partnership index is based on a logistic regression model. Model parameters are estimated and its accuracy is tested on partnership data from the Estonian Labour Force Survey and the Estonian Social Survey. To avoid assigning one person to multiple households, the partnership data is considered as a bipartite graph and stable marriage matching is calculated with Gale-Shapley algorithm. The model accuracy is currently 86%. The work on improving the index continues.
ISSN:1874-7655
1875-9254
DOI:10.3233/SJI-180484