Preparing IPUMS samples for Ireland. Deirdre Cullen Senior Statistican

Transcription

1 Preparing IPUMS samples for Ireland Deirdre Cullen Senior Statistican

2 The History of Census in Ireland the first full 6census of Ireland The so-called Great Census of 1841 first modern census of Ireland 6.5m Persons 5 4 The Great Famine no census because of War of Independence and Civil War First population 3 estimates in m 2 1 Censuses at 10 year intervals up to 1911 under the British system of administration 3.1m 1926 first census taken in new independent Ireland 3.0 Persons Since 1951 a census every 5 years

3 Past census micro-data files Samples of anonymised records (SAR) were produced for 1996, 2002 and 2006 census Samples were 5% of person records and were held by University College Dublin Data was top-coded Access to samples was through university

4 IPUMS for Ireland First meeting with Robert McCaa in May 2009 Census data is available in electronic file format for 8 censuses over period Agreed to supply samples and metadata for all 8 Census years 2002 and 2006 data to be delivered by end of 2009 Remaining data to follow in 2010

5 Preparation of samples: Planning IPUMS would take unit record files and do all necessary work on them Statistical confidentiality policy of CSO made this impossible Risk of statistical disclosure through release of detailed records contravenes Statistics Act 1993 Also risk of being gperceived to have divulged sensitive information to a foreign agency Therefore all anonymisation and sampling done by Therefore all anonymisation and sampling done by CSO

6 Preparation of samples: Data Samples produced using SAS SAS code available for SAR for 2006, 2002 and used as starting point SAR code was for 5% of persons needed to modify for 10% household sample 2 tasks to accomplish: Extract 10% sample of households Top-code sensitive variables to avoid any disclosure

7 Variables Number of variables: no household variables no household variables mini census to replace cancelled

8 2006 variables - Person Age, sex, marital status, place of birth, place of usual residence, nationality, ethnicity, religion Family composition, disability, carers, voluntary activities Iihl Irish language (ability and frequency), education (level and field) i l i Economic status, employment status, occupation, industry, socio-economic group and social class

9 2006 variables - dwellings Type of accommodation, year built, nature of occupancy, Heating (y or n), type water supply, type sewerage Number of cars, PC (yes or no), internet access

10 Top-coding In addition to random element, top-coding of all variables was carried out Because complete households h were included dinstead of individuals, age was now a more sensitive variable (age structure within household could lead to identification of large families etc.) Age was given in 5-year groups for persons aged 20+ and SYOA for aged <20

11 Top coping Geography regional authority areas

12 Top-coding Age - 5 yr groups for aged 20+ Marital itlstatus tt - Collapsed d7t to 4 Level of education - Collapsed 13 to 5 categories Place of birth, UR - Full detail Because regional Industry - Full detail authority area only Occupation - Full detail Religion - Collapsed to 4 categories Ethnicity - Collapsed from 7 to 3 categories Voluntary activity - Collapsed 5 to 2 categories yes or no Disabiltiy - Collapsed 9 to 2 categories- yes or no Carer - Yes or no Area type (townsize) - Collapsed 15 to 7 categories Year of return - Coded to 7 values Water - Collapsed 5 categories to 4 Sewerage - Collapsed 5 categories to 4

13 Metadata - documentation Most metadata requested by IPUMS was available Exceptions technical & analytical reports and post enumeration surveys (not ( t carried i d out) t) Much of the older documentation was in hard copy only had to be photocopied and sent by mail Some metadata not available, but was created during production of samples ((e.g. g Country y classifications)

14 Metadata - classifications Record layout available for all files Lack of detailed classifications for old files No places classifications available, others changed over time Had to recreate published data in order to derive classification used Bucket groups tricky Meant we were limited to what had been published

15 Outcome December and 2002 sent March and 2002 resent, 1996 sent May , 1986, 1981, 1979, 1971 I l d h l b f C Ireland now has largest number of Census years on IPUMS website

16 Resources 1 Statistician 6 months 1 line manager small input (2 weeks) 1 staff 2 weeks

17 Advantages of IPUMS for Ireland Bonus for CSO: as a result of this project, our historic data sets are now in a much more usable format IPUMS allows mix of Census years available in 1 file Comparability with other countries Ease of access for users Positive publicity for Census in Ireland

18 Assessment of IPUMS website Website is excellent in terms of ease of navigation and user-friendliness FAQ and documentation on site is clear and useful Shopping-cart style system for selecting variables works very well Instantly shows which variables are available across Instantly shows which variables are available across all samples selected

19 Assessment of IPUMS samples Extract download was very efficient Included code to input the data into a SAS data set very useful: included all format statements etc. All years and countries selected go into 1 file very useful lfor comparisons Of use even within CSO to do basic comparisons with other countries to inform thinking on more detailed analyses

20 Questions