2. Outline
1. What is LINKS?
2. The pipeline
• Civil registry
• Data entry
• Linking the certificates
• Family reconstitutions
3. Performance / Experiences thus far
3. General idea
Vital events that occurred after 1811:
• have been registered.
• are (or will be) indexed and digitized
• can be linked with LINKS
• which is necessary to reconstitute families
In time, this makes it possible to retrace every individual
who experienced a vital event in the Netherlands.
4. 1. What is LINKS?
Family reconstitution using birth, marriage, and death
certificates from the Dutch civil registry
LINKing System for family reconstitution
Contains information on:
• parents, offspring, and spouses,
• age at marriage, childbirth, and death,
• number of marriages, children, etc.
• birth intervals,
• migration
• occupations
8. 2a. Civil registry
Starts in 1796/1811/1812
Third oldest in the world after France (1796) and Belgium (1796/1811)
Contains vital events
• Birth certificates 1811-1918 100-year delay
• Marriage certificates 1811-1943 075-year delay
• Death certificates 1811-1968 050-year delay
Registered at the municipal level
• Made in duplicate
• No loose forms
• Pages watermarked by local judiciary
Civil registry
10. Useful for research
Information:
• Vital events (duh)
• Municipality of registration
• Family relations
• Age
• Occupations (not always indexed)
Available certificates:
• Birth certificates 14,215,411
• Marriage certificates 04,642,000
• Death certificates 13,212,269
F1
F2
F3
First marriage Second marriage
11. Unique selling points
Families!!!
• Clustering
• Networks / capital
• Intergenerational transfers
Size:
• 2+ generations
• Complete record of families
• Ample half-sibs, second cousins, etc.
• Large database
• Sizable area
F1
F2
F3
First marriage Second marriage
13. 2b. Data entry
Regional archives make indexes of certificates:
Choose themselves what to index:
• Occupations,
• Witnesses,
• Names parents (now entered)
• Varies by region
CBG/Genlias/WieWasWie:
gather/own data and have contract with IISG for LINKS
Civil registry
Indexed
registry
Data entry
14. Graphic summary by Kees Mandemakers
‘WieWasWie’
database/
website
ARCHIVES
archives
including CBG,
Digital Family Tree
LINKS
cleaned
Feedback
15. Data entered thus far
Done:
Drenthe/Groningen, Flevoland (Urk, Schokland), Gelderland, Limburg,
Noord-Brabant, Overijssel, Utrecht, Zeeland
Omissions:
Friesland, Noord-Holland (A’dam excluded), Zuid-Holland
Sources:
WieWasWie/CBG
IISG Amsterdam
Bob Coret
17. 2c. Linking the certificates
Matching birth, marriage and death records
Life course
1. Birth - Marriage
2. Birth - Death (children)
3. Marriage - Death (married, no birth)
Intergenerational
4. Birth - Marriage parents (best)
5. Marriage - Marriage parents (good)
6. Death - Marriage parents (limited usefulness)
Cleaned
registry
LINKS
Linking
18. Linking the certificates
Name constraints:
• Fixed first syllable
• Exact matching
• Levenshtein dependent on length of name
Time constraints:
• Age at death
• Age at marriage / between ages 15-50
22. Results
Match Result Selection
Intragenerational
Birth – Death 67.5% Born: 1812-1862
- post hoc matching 68.8% Born: 1812-1862
Marriage – Birth 85.7% Married: 1862-1912
Marriage – Death 68.2% Married: 1812-1862
- post hoc matching 68.5% Married: 1812-1862
Intergenerational
Birth – Marriage parents 82.4% Born: 1842-1912
Marriage – Marriage parents 82.5% Married: 1887-1937
23. Some thoughts on matching
• Birth and marriage certificates are most easily
matched (required at marriage).
• Death certificates are harder to match.
• Hence, birth and marriage registers should be the
basis to reconstruct the population of the
Netherlands between 1812-1968.
• However, death certificates need to be matched,
so that end users can use the data.
24. Known problems
Source:
• Double entry of registration
• Age missing / spelling variations (also during entry)
• Legal reality, rather than biological
- No extramarital family relations
- Real father?
Matching:
• Migration
• Premarital births
• Twins
• Reused name within family
25. 2d. Family reconstitution
Product is linked certificates
But users require reconstituted life courses or families!
Problems to solve:
• Make analyzable structure
• Links should be unambiguous (same parent identified)
• Unmatched certificates within reconstituted families
No standardized operationalization (yet)
LINKS Dataset
Reconstitution
31. 3. Performance / Experiences thus far
Used to publish in top journals (Nature Communications)
Structure is tricky. Matches are good, but the questions is
what we miss and why.
Observations are only useful if:
• A death certificate is available
• A marriage certificate of the parents is available
Publish R-script + manual together with Ingrid van Dijk to
transform LINKS into a dataset.
32. Take-home message
1. A pipeline already exists to move from raw data to
LINKS-Zeeland.
2. Data is of ample quality to publish in high impact
journals.
3. Full civil registry of the Netherlands is available in
the archives
• However, not all certificates have been entered yet
• Not all variables are indexed
4. Question is how to use the available infrastructure