editing census data: mexico’s experience

23
Editing Census Data: Mexico’s Experience Oswaldo Palma INEGI, Mexico September 2012

Upload: newman

Post on 14-Feb-2016

54 views

Category:

Documents


0 download

DESCRIPTION

Editing Census Data: Mexico’s Experience. Oswaldo Palma INEGI, Mexico September 2012. About INEGI. National Institute for Statistics and Geography (INEGI) is, from 2008, an autonomous institute in Technical and Managing matters. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Editing Census  Data: Mexico’s Experience

Editing Census Data:Mexico’s Experience

Oswaldo PalmaINEGI, MexicoSeptember 2012

Page 2: Editing Census  Data: Mexico’s Experience

About INEGI

National Institute for Statistics and Geography (INEGI) is, from 2008, an autonomous institute in Technical and Managing matters. According to Mexican Statistical and Geographical Information Act, INEGI it is the responsible for making the Population and Housing Census every decadeBeing the responsible for the Statistics-Geography couple, it gives institute significant technical, operational and statistical data dissemination advantages.

Page 3: Editing Census  Data: Mexico’s Experience

Characteristics of Census 2010

Traditional census, de facto.Field working to visit around 32 million of dwellings and enumerate 112.3 million of inhabitants.Field work in the Mexican 2010 Population and Housing Census was undertaken from May 31st to June 30th, 2010.Interviews were held using printed forms.

Page 4: Editing Census  Data: Mexico’s Experience

Forms used Six kind of forms were used:• Building list• Short form• Long form• Self enumeration form• Urban environment form (CEU)• Locality form

Page 5: Editing Census  Data: Mexico’s Experience

Information’s TreatmentDelivering results in eight monthsParallel processesSix stages• Form’s storage• Data Entry• Coding• Editing• Mapping update• Figures validation• Disseminating

Page 6: Editing Census  Data: Mexico’s Experience

Editing Short form and Long form

Page 7: Editing Census  Data: Mexico’s Experience

Data Editing characteristics

Assure the logic congruence and integrity of questions and sectionsCorrecting errors made during the interview or data entryRespect as much as possible the respondent answersPractically null stochastic imputation

Page 8: Editing Census  Data: Mexico’s Experience

Methodology implemented

Vector’s methodolgy was used• To create an equivalence between cases (a

partition)Advantages• Comprehensive information analysis• System programming simplified• Analysis simplified• Follow-up simplified on changes made

Page 9: Editing Census  Data: Mexico’s Experience

Vector’s Methodology stepsVariables intervening in editing criteria and values taken for each variable are identified.A vector is built which components represent values that every variable can take.All combinations between values that vectors take are generated, which are obtained by varying every value of each vector component, this is made starting from last to first component up to obtain the total group of combinations.To have control on combinations generated, the known addressing function is built, which allows the assignment of an only value also known as image to each combination according to the order in the one they are generated.For each particular combination an specific treatment is stated.

Page 10: Editing Census  Data: Mexico’s Experience

Number of criteria designed232 criteria were programmed using JAVA supported by PL/SQL inside database

Section Total of Criteria

Initial Treatments 7 Dwelling Characteristics 29 Food Access 3 Persons Data 6 Sex and Age 8 Relationship 35 Age Cut-off 8 Entity or Country of birth 4 Health Services 4 Religion 2 Disability 15 Indigenous Language 4 Educational Characteristics 12 Residence Entity or Country in 2005 8 Marital Status 10 Economic Characteristics 30 Fertility 27 International Migration 20

232 Each criteria was tested using a “Editing Criteria

Simulator”

Page 11: Editing Census  Data: Mexico’s Experience

Computing infrastructure

Robust architecture Supporting multi-processingPossibility to add Apps & DB servers according to requirementsTime to process 165.6 hours

3 Apps Servers SERVER A

Containers Virtual machines Memory per Machine OC4J_CPV2010_CTRL 1 512MB OC4J_CPV2010_CARDES 2 2GB OC4J_CPV2010_VALIDA_A 2 2GB SERVER B

Containers Virtual machines Memory per Machine OC4J_CPV2010_VALIDA_B 2 2GB OC4J_CPV2010_MEDAM 1 1GB SERVER C

Containers Virtual machines Memory per Machine OC4J_CPV2010_VALIDA_C 2 2GB OC4J_CPV2010_REPORT 1 1GB

3 Oracle Apps: App Resources

Val1Pob2010 Memory 3GB Disk space 50GB

Val2Pob2010 Memory 3GB Disk space 50GB

Val3Pob2010 Memory 3GB Disk space 50GB

Page 12: Editing Census  Data: Mexico’s Experience

Editing Urban Enviroment form

Page 13: Editing Census  Data: Mexico’s Experience

Urban Environment form (CEU)

CEU was carried out for first time in 2010Allows the collection of information about urban characteristics of streets delimiting each block in localities of 5,000 inhabitants or overInclusion of CEU in the 2010 Census forms allows link information on population and housing with urban characteristics.

Page 14: Editing Census  Data: Mexico’s Experience

Characteristic collectedKind of road of characteristic (street, avenue, etc.)Persons access (free, restricted, prohibited)Cars access (free, restricted, prohibited)Type of recovering (concrete, stone, ground)Availability of street identifier Availability of:• Public light.• Public telephone.• Sewage.• Sidewalk.• Sidewalk margin.• Ornamental trees or plants.• Collective transport.• Semi-fix trade workstations.• Petty trade workstations.

Page 15: Editing Census  Data: Mexico’s Experience

New Challenges

A need to generate vector mapping for new geographic levels (block side).Generate and maintain a new conceptual framework consistent to prior traditionally taken into account.Develop new methodologies to: coding, editing, validating, and imputing to ensure information consistency when this is represented on a map.

Page 16: Editing Census  Data: Mexico’s Experience

Example of inconsistencies

Page 17: Editing Census  Data: Mexico’s Experience

Problem solving advance

Information association to its respective block sideStandarization of streets namesIdentification of irregular patterns in mapping

Page 18: Editing Census  Data: Mexico’s Experience

Association and Standardization

100 analysts hired during four months

Page 19: Editing Census  Data: Mexico’s Experience

Main problems foundMethodologies and concepts to unify language and strategies are required.Training staff in conceptual model (long time process)Hiring or training of human resources in handling geographic information apps. (It could be an issue specially when the NSO don't conjugates the statistical & geographical expertise)Development of expert systems which allows automatic and assisted identification of patterns and data infrastructure needed for these activities.Computers with proper capacity and performanceSometimes bandwidth can also be an issue

Page 20: Editing Census  Data: Mexico’s Experience

Administrative registersNowadays, the first essay to associate the administrative registers of electricity supplier with the cartographic frame is being undertaken. In this process the information and methods derived from the standardization process of street names is used as key information. (72% of the essay information has been associated).However; these tests, are oriented to define a proper conceptual framework incorporating linguistic computing elements and pattern recognition in texts.

Page 21: Editing Census  Data: Mexico’s Experience

Concluding remarks

Data editing of the Census 2010 was successfully carried out.• Significant improvement compared to prior

census were obtained.• Improvements are mainly identified in the

changes control on information and treatments comprehensiveness.

• The time to carry out census data editing was almost a year (lesser than the previous censuses)

Page 22: Editing Census  Data: Mexico’s Experience

Concluding remarks

Editing CEU had no problems when considering it as the traditional editing registers process in a data table. The development of criteria considering data consistency when these are presented on a map still is a challenge. • Latter imply interesting technical challenges and

can force to the NSO to review planning and evaluate available resources.

Page 23: Editing Census  Data: Mexico’s Experience

Concluding remarks

Finally, for the case of Mexico, which has no administrative registers with the quality needed for use as statistical registers, work for solving the problem of data editing can be reused to convert the available registers in statistical registers.