IT in the Brazilian 2010 Census: New Technologies in Population and Housing Censuses Roberto Luis Olinto Ranos Director of Surveys

Roberto Neves Sant’Anna Head of International Relations

46th Session of the UNSC New York, USA, 27 February 2014

Brazil 2010 Census showcase Numbers Area: 8.5 millions Km2 Municipalities: 5,565 Households: 58 million Habitants: 190,732,694 Enumeration areas: 314 thousand Enumerators: 191 thousand Mobile devices: 150,000 smart phones and 70,000 PDAs Collection Offices: 6,823 Budget: R$1.67 billion (67% staff, 19% current expendures and 14% investments) Source: IBGE

Main Goals Ensure geoprocessing of data • •

Control coverage area of collection and guide the route of enumerator New forms of data dissemination, allowing the integration of statistical and spatial information

Speed up survey data processing • • • •

Guarantee completion of mandatory items (without returning to the field) Simultaneous respondent filling Reduce steps Allow real-time monitoring of data collection

Guarantee Data Security • •

Encrypted data Daily transmission

Guarantee Data Quality • •

Minimize filing errors or failure to complete mandatory items Allow real-time monitoring of data quality

Reduce flow of paper National register of addresses for statistical purposes (CNEFE)

Brazil 2010 Census showcase Data Collection Model IBGE Offices with desktop software JAVA

Intranet

PDA with no data

Download Software Input data

Upload Central

Internet questionnaire

Output data

PDA with data

Brazil 2010 Census showcase Background of using mobile devices for data collection 2007 Census

2007 National Household Survey

2010 Census

Internet Connection

Dialup Satellite 2G and 3G

Dialup Satellite 2G and 3G

Satellite 2G and 3G

Devices

PDAs MIO 550b with Windows mobile 5 No Wi-Fi

PDAs MIO 550b with Windows mobile 5 No Wi-Fi

LG Sm@rt GM750 with Windows Mobile 6.5 and PDAs MIO 550b with Windows mobile 5 No Wi-Fi

Questionnaire Application

Developed internally using proprietary software

Open source SDK

Open source SDK

Additional Data Collection Mode

Internet data collection

Georeferenced Data Collection

Challenges of data collection Lack of electricity, lack of internet access

Exposure to bright sun light

Rural areas with difficult access and jungle

Addressing the challenges • • • •

Battery charging – taking extra charged batteries Transferring of information Poor visibility under bright sunlight Heterogeneity of communication services with slow internet speed in many municipalities due to cost and difficulty of installing satellite antennas

Challenges of data collection Respondent not available or has no time to answer

Addressing the challenge •

Internet option integrated

Achievements Time before dissemination 1991 Census

2000 Census

2010 Census

Number of questionnaires

35 million basic questionnaires (BQ) 4.5 million sample questionnaires (SQ)

42 million BQ 5.5 million SQ

58 million

Methodology used to digitalize data

1200 typists

Use of scanners with bar code and mark recognition (OMR), printed characters (OCR) and manuscripts (ICR) on the image of questionnaires

50,000 smart phones and 70,000 PDAs

Collection timeframe

3 months

3 months

3 months (including analysis)

Digitalization

6 months for BQ 8 months for SQ

4 months

--

Infrastructure

21 centers of data entry

NA

6,823 collection offices

Source: ENAP Available at: http://inovacao.enap.gov.br/index.php?option=com_docman&task=doc_view&gid=277

Different dissemination methods

Advances in the 2010 Census Use of satellite images • The use of high resolution satellite images allowed considerable progress in

identifying and delimiting subnormal agglomerates;

Macapá (AP)

Advances in the 2010 Census Use of satellite images

Belo Horizonte (MG)

Advances in the 2010 Census

Association of Census data to risk areas

The BATER is a territorial cut defined specifically for association of the 2010 Population Census data to existing risk areas. It consists of the smallest possible territorial cutout for dissemination of census data in

areas of risk, and may be composed of smaller areas than an enumeration area. For example, a set of block faces, in compliance with the statistical confidentiality conditions.

Construction of BATER used the advances in the IBGE address register, using the block face as the main territorial unit to its composition.

Advances in the 2010 Census Exemple of a BATER

Future data dissemination Statistical Grid Dissemination Hierarchical spatial structures formed by regular cells and used to make aggregated data available

Future data dissemination Municipality Dissemination

x

1:8,000 ~4.6 km

Grid Dissemination

1:2,000 ~1.1 km

Future data dissemination selected area

enumeration area as smallest part

block face as smallest part

Conclusions •

The mobile devices brought agility to the whole survey process with a guarantee of data quality and security

•

Improved management of fieldwork

•

Avoided the transport of large volumes of paper questionnaires

•

Allowed new forms of data dissemination

•

The combination with internet allowed to get data from closed households

•

How can we improve future census data collection? Multi modal? (PDA, Internet, etc.)

Sharing the experience with other countries - data revolution in practice National Statistical Institutes of Africa and South America that have adopted, in the round of the 2010 Census, the same technology and information capture process: • • • • • •

Cape Verde Sao Tome and Principe Senegal Ivory Coast Paraguay Uruguay

Thank you!

roberto. [email protected] [email protected]

www.ibge.gov.br