The Belgian HISSTAT project

Transcription

1 The Belgian HISSTAT project Documenting and reconstructing the 1961 census sample Wouter Ronsijn Free University of Brussels (VUB) Session Data Management and Data Analysis in Quantitative Historical Social Research (Dr Ronald Gebauer, Dr Axel Salheiser) Fifth Conference of the European Survey Research Association, Ljubljana, Slovenia, July 15-19, 2013

2 The Belgian HISSTAT-project ( ) Recent years: development of large-scale historical databases Belgium: HISSTAT-project: to protect and exploit Belgium s rich statistical heritage Data since 1800: beginning of uniform administrative structure (French reforms) beginning of uniform data-collection for whole territory After 1830: Belgian statistical administration Led by people such as Adolphe Quetelet 1846: First large-scale population, agricultural and industrial census

3 The Belgian HISSTAT-project ( ) Historical census results: communicated in bulky publication, often several volumes 1866 population census

4 The Belgian HISSTAT-project ( ) Goal of HISSTAT: to digitise these data: 1/ Cross-sectional aggregated data Basis: municipality Since: 1800 Content: official censuses (population, agriculture, industry); other data: taxation, elections, cadastral data, 2/ Cross-sectional microdata Basis: individual Since: 1961 Content: population censuses, linked to other records 3/ Aggregated longitudinal data Basis: municipality Since: 1880 Content: population, births, deaths, migration

5 The Belgian HISSTAT-project ( ) GIS-application: possible to display data on historical maps at the level of the municipality Population density in Belgium, 1846

6 The Belgian HISSTAT-project ( ) HISSTAT-website and online applications: general project website municipal-level population, agricultural and industrial census data ± municipal-level demographic and socioeconomic data from five successive population censuses,

7 The 1961 census Census carried out on December 31 st, 1961 ± census forms collected Census results processed by means of punched cards 1 card for 1 individual, 2 nd card for individuals with diploma(s) Individual punched card for 1961 census

8 The 1961 census sample file Basis of sample file: punched cards Donated to BASS (Belgian Archives for the Social Sciences, at the University Catholique de Louvain-la-Neuve) around 1970 Selection of 10 % read to magnetic tape around 1970; copied to hard drive around K P 00 & & } } & & Q & M P 00 & J R5801 & & R R & Q & & N L 00 & & N & & & M }4600 & & & & & J & M O 00 &

9 The 1961 census sample file Reconstruction of codebook Some codes still unexplained (presumably typing or reading errors) Recoded to numeric characters only (original codes also had alphabetic characters) Problem with years: distinction between 19th and 20th century, e.g.: 55 = N = 1855 What with e.g. 77 (= 1977)?

10 The 1961 census sample file Problem with data on diploma s: Cases with data at the correct position:...columns columns columns columns & L M5400 & O & Cases with data at the wrong position:...columns columns columns columns & & &

11 The 1961 census sample file Representativity of the sample: 10 % of the population? No information at household level: sample of individuals, not of households Unkown how the sample was made Final version of the file: cases

12 The 1961 census sample file Representativity of the sample: 10 % of the population? Antwerp Ghent Brussels Liège Charleroi Namur Sample fractions per arrondissement Total population 8.5 to 9.5 % Total population 9.5 to 9.9 % Total population 9.9 to 10.1 % Female population < 9.8 % Male population < 9.8 %

13 The 1961 census sample file Representativity of the sample: 10 % of the population? Antwerp Ghent Brussels Liège Charleroi Namur Gender balance deviation: difference from published results in number of men per 100 women Deviation from gender balance in the sample, by arrondissement [-10 to ]-3 [-3 to ]-1 [-1 to ]1 [1 to ]3 [3 to ]4

14 The 1961 census sample file Representativity of the sample: 10 % of the population? Three problematic arrondissements: result of data loss: 1/ Loss of data when data were copied from tape to hard drive (arrondissement of Huy) 2/ Loss of punched cards arrondissement of Sint-Niklaas: 3 municipalities have almost no men arrondissement of Kortrijk: town Menen: half of the women are missing Except for these problematic arrondissements: sample more or less representative at the level of arrondissements

15 Conclusions 1961 census sample: oldest microdata available for the whole Belgian population File has some shortcomings: not at household-level, loss of data Still: shortcomings should not stand in the way of using it for scientific research HISSTAT team hopes these and other data it has gathered will open new perspectives for scientific research Contact HISSTAT: (Free University of Brussels) (Ghent University)