1960 Dutch Population Census First digital census with

1960 Dutch Population Census • First digital census …with analogue problems

1960 Census • 11 million depersonalised individuals on 11 million punch cards • 2000 cards per box, 5500 boxes • Stored in at least two locations • Read in 1973, stored on tape • Briefly looked at in 1982 • Analysed and cleaned in 1994

1960 Census Aims: • Investigation of flaws in the current dataset and proposals for correction and reconstruction • Proposal for publication in three ways: 1. individual data under strict conditions 2. aggregated data supplemented with published tabels 3. sample for IPUMS-international

1960 Census Investigation of flaws • About 300. 000 records are missing – 24 municipalities in 4 provinces are missing – in about 450 municipalities 1 -10 men and/or women are missing – at best 3000 records may be reconstructed from previously eliminated records • About 250. 000 records are superfluous – few municipalities have (partly) been read twice

1960 Census What happened? • Some punch cards could not be read • Some punch cards were demolished whilst being read • Some boxes with punch cards were left out • Some boxes with punch cards were inserted twice • Some records had faulty values and were eliminated • Time pressure was high • There is NO documentation

1960 Census • Hollerith tabulator(1890 US Census)

1960 Census Punch card as used in the highly contested US Presidential elections in 2000

1960 Census Example of eliminated records 1115100421115100421110 1115100421116 1115100421119 121586010 12158 W 000 121586001 6302120001000995581111405057126086200 1306363301000075 -81111718035817732405 1305352202000900521111205041728284204 4303430001000930521111203038829276500 3855413012060 3755113010010 3406713012050 < seems to represent 1 S seems to represent 2 W seems to represent 6 B"("N 3=")"5 ZD, 10 B SC 2+NSC 3); ", /")"); B"("N 3=")"5 ZD, 10 B 3 52701322981010 2 52801322981010 00 0152701322981010 A seems to represent 1 V seems to represent 5 Z seems to represent 9 1760 1770 1780 1790 1060 0061 1860

1960 Census What happened? • Some punch cards could not be read • Some punch cards were demolished whilst being read • Some boxes with punch cards were left out • Some boxes with punch cards were inserted twice • Some records had faulty values and were eliminated • Time pressure was high • There is NO documentation

1960 Census Proposals for reconstruction • How many flaws may a census have and still qualify as a proper census as opposed to an extensive survey? • What standards should it meet? Data entry from published tables Weighing factors Fictional individual records derived from tables?

1960 Census Common flaws • Some (how many? ) records get lost • Non-response • Mistakes in counting How to deal with these flaws? Are they any different from our current problems?

1960 Census Proposals for reconstruction • How many flaws may a census have and still qualify as a proper census as opposed to an extensive survey? • What standards should it meet? Have/can standards been defined? Data entry from published tables Weighing factors Fictional individual records derived from tables?

1960 Census How much effort should the reconstruction of 4587 missing Rotterdammers (0. 62%) take?

1960 Census Many small municipalities render much data-entry for the sake of few people.

1960 Census Proposals for reconstruction · Up to 100 records and less than 1% of records missing · 100 to 1000 records and 1% to 5% of records missing • More than 1000 or more than 5% of records missing

1960 Census Proposals for reconstruction • How many flaws may a census have and still qualify as a proper census as opposed to an extensive survey? • What standards should it meet? Have/can standards been defined? Data entry from published tables Weighing factors Fictional individual records derived from tables?

1960 Census • • Aims: Investigation of flaws in the current dataset and proposals for correction and reconstruction Proposal for publication in three ways:

1960 Census • • Aims: Investigation of flaws in the current dataset and proposals for correction and reconstruction Proposal for publication in three ways: 1. sample for IPUMS-international

1960 Census Aims: Investigation of flaws in the current dataset and proposals for correction and reconstruction Proposal for publication in three ways: • • aggregated data supplemented with published tabels sample for IPUMS-international

1960 Census Aims: Investigation of flaws in the current dataset and proposals for correction and reconstruction Proposal for publication in three ways: • • • individual data under strict conditions aggregated data supplemented with published tabels sample for IPUMS-international

1960 Census • Central Statistics will monitor the output • There must be a satisfactory solution for the missing records