steering committee of the international monitoring ...bhas.ba/census/1348896-report 21 - sc mission...

Steering Committee of the International

Monitoring Operation on the Population and

Housing Censuses in Bosnia and Herzegovina

Twenty-first Assessment Report

24 – 27 February 2015

Jean-Michel DURR, Senior expert, SC Team leader

Meryem DEMIRCI, GUIDO PIERACCINI, Experts

International Monitoring Operation on the Population and Housing Census in Bosnia and Herzegovina 2

Contents

LIST OF ACRONYMS ........................................................................................................................................ 3

INTRODUCTION AND BACKGROUND ....................................................................................................... 4

EXECUTIVE SUMMARY .................................................................................................................................. 7

MANAGEMENT AND DECISION MAKING PROCESS .............................................................................. 8 Recommendations ........................................................................................................................................................... 8

DATA PROCESSING ......................................................................................................................................... 9 CODING .............................................................................................................................................................................. 9

Assessment .......................................................................................................................................................................... 9 Recommendations ........................................................................................................................................................ 11

DATA EDITING ............................................................................................................................................................... 11 Assessment ....................................................................................................................................................................... 12 Recommendations ........................................................................................................................................................ 12

PERSONAL INFORMATION AND CENSUS MATERIALS ...................................................................... 13 Recommendations ........................................................................................................................................................ 13

POST ENUMERATION SURVEY .................................................................................................................. 13 Assessment ....................................................................................................................................................................... 14 Recommendations ........................................................................................................................................................ 14

DISSEMINATION OF CENSUS RESULTS .................................................................................................. 15 Assessment ....................................................................................................................................................................... 15 Recommendations ........................................................................................................................................................ 15

ANNEXE ............................................................................................................................................................ 16 AGENDA OF THE MISSION ............................................................................................................................................ 16


List of acronyms

BiH Bosnia and Herzegovina

BHAS Agency for Statistics of Bosnia and Herzegovina

CoM Council of Ministers of BiH

CMIS Central Monitoring Information System

CSO Civil Society Organisation

DB District Brčko

EA Enumeration area

FBiH Federation of Bosnia and Herzegovina

FOS Federal Office of Statistics (FBiH)

EI Entity Instructor

EN Enumerator

EUPHC 2 Technical assistance to Population and Housing Census Phase II

GIS Geographic Information System

ICEI Department for International Cooperation and External Information

IMO International Monitoring Operation

IMO MG International Monitoring Operation, Management Group

IMO SC International Monitoring Operation, Steering Committee

IT Information Technology

LFS Labour force survey

MCC Municipal Census Commission (Census Commission of the units of local self-

government)

MI Municipal Instructor

PES Post Enumeration Survey

Q Question

RS Republika Srpska

RSIS Republic Srpska Institute for Statistics

SA Statistical Area

SI Statistical Institution

TA Technical Assistance

TAP Technical Assistance Project

TOR Terms of Reference


Introduction and Background

1. A Population and Housing Census shall take place in Bosnia and Herzegovina (BiH)

in accordance with the Law on the Census of the Population, Households and Dwellings in

Bosnia and Herzegovina in 2013 as adopted by the Parliamentary Assembly of Bosnia and

Herzegovina on 3 February 2012. The Council of Ministers of Bosnia and Herzegovina has

invited the European Commission to organise the international monitoring of the Census.

Therefore, the European Commission, the Council of Europe and the Council of Ministers of

Bosnia and Herzegovina signed on 18 April 2012 a Memorandum of Understanding to agree

upon the following:

- The general objective of the International Monitoring Operation (IMO) of the Population

and Housing Census in Bosnia and Herzegovina is to monitor the compliance of the

whole Census exercise, from the preparation to the data dissemination, with:

- International standards on population and housing censuses as defined by UNECE and

Eurostat, and as adopted by the Conference of European Statisticians as Recommenda-

tions for the 2010 Censuses of Population and Housing;

- Regulation (EC) No 763/2008 on population and housing censuses, and its implementing

measures;

- The Fundamental Principles of Official Statistics, adopted by the UN Statistical Commis-

sion, as well as the European Statistics Code of Practice, promulgated by the European

Commission;

- Standards on data protection and confidentiality, as provided for in the Convention for the

Protection of Individuals with regard to Automatic Processing of Personal Data of the

Council of Europe, and the relevant European Union regulations in force.

2. The Operation will be carried out by a Committee of International Organisations, the

Management Group (IMO MG) assisted by a Senior Census Expert, experts in population

censuses, one or more experts in information technology and persons monitoring the census

enumeration in the field.

3. The Senior Census Expert, the Census Experts and the IT Experts will assess the

compliance of the Census in Bosnia and Herzegovina with the requirements listed in the

Memorandum of Understanding, but they will not provide technical assistance.

4. Under the guidance of the Committee and the Senior Census Expert, the Census Ex-

perts shall:

- Examine the technical preparation of the Census including the pilot Census and post-

enumeration survey, in particular the drafting of questionnaires and manuals;

- Monitor the collection, processing and dissemination of Census data, verifying the ac-

cordance with the requirements defined above;

- Control the fair and proper computation at all levels and whether the confidential nature

of the individual data is fully guaranteed;

- Investigate any other matter with relevance for the Census.

5. The Management Group has designated Mr. Jean-Michel Durr as Senior Census Ex-

pert and Mr. Roberto Bianchini, Mr. Bent Noerby Bonde, Ms. Meryem Demirci, Ms. Kateri-


na Kostadinova-Daskalovska, and Mr. Guido Pieraccini as experts in population censuses.

The Senior Census Expert and the Experts in population censuses form the IMO Steering

Committee (IMO SC) of international experts.

6. The first mission of the IMO SC was conducted from 23 to 26 April 2012. The objec-

tive of this mission was to assess the preparations for the population and housing census in

their main components, including the preparation of the pilot census. The report of the SC

was sent by the Senior Census Manager on May 12 to the IMO MG.

7. The second mission was conducted from 26 to 29 of June 2012. The objective of this

mission was to assess the progress made since the first mission in preparations for the popula-

tion and housing census, including the preparation of the pilot census.

8. The third mission took place on 17 and 18 of September 2012 and was conducted by

two members of the SC, namely, Mr. Jean-Michel Durr and Ms. Katerina Kostadinova-

Daskalovska. The objective of the mission was to assess the implementation of recommenda-

tions made in previous reports and to assess the preparation of the pilot census, planned for

October 15-29.

9. The fourth mission took place between 10 and 29 of October 2012 and was conducted

successively by the members of the SC (5 days each, except the senior expert from 10 to 19

of October). The objective of the mission was to observe the conduct of the pilot census in

the field. In addition, a team of four observers, composed of Lidija Naumovska, Gabor

Rosza, Jean-Paul Sardon and Per Schöning was present during the whole period of the pilot

census. Each of them covered several municipalities, in order to observe the pilot census field

operations in all the 60 enumeration areas (EAs) sampled in the pilot.

10. The fifth mission took place from 19 to 21 of November 2012 (3 days) and was con-

ducted by one member of the SC, Ms. Katerina Kostadinova-Daskalovska. The objective of

the mission was to observe the conduct of the Post enumeration survey (PES) of the Pilot

census in the field. The expert covered all 6 EAs (in 6 municipalities) sampled in PES sample

in order to observe the PES field operations.

11. The sixth mission took place from 10 to 13 of December 2012 and was conducted by

four members of the SC: Mr. Roberto Bianchini, Mr. Bent Noerby Bonde, Ms. Katerina

Kostadinova-Daskalovska and Mr. Guido Pieraccini. The objective of the mission was to hear

from the Statistical Institutions about the lessons learn from the Pilot Census and to assess

whether the preparations are on level that could allow conducting a census in April 2013.

12. The seventh mission took place from 12 to 15 February 2013 and was conducted by

one member of the SC, Mr. Roberto Bianchini. The objective of the mission was to assess the

situation on the progress of preparation and implementation of the IMO recommendations in

the new context of the postponement of the census, with a focus on the achievement of the

deadlines regarding cartographic preparations. A second objective was to collect information

and materials for a more detailed assessment on the progress of census preparatory activities

to be made by all members of the SC in March 2013.

13. The eighth mission was conducted by the complete SC from 18 to 22 March 2013.

The objective of the mission was to assess in detail progress made in the preparations with

regard to the milestones given after the sixth mission in December 2012, when the SC rec-

ommended to postpone the census to October 2013.


14. The ninth mission was conducted by two members of the SC, Jean-Michel Durr and

Katerina Kostadinova-Daskalovska from 22 to 25 April 2013. The objective of the mission

was to assess the progress made in the preparations with regard to the milestones given in the

sixth mission and updated in previous missions.

15. The tenth mission was conducted by the full SC from 28 to 31 May 2013. The objec-

tive of the mission was to assess whether the progress made in the preparations was sufficient

to consider that the country was ready to conduct a census in October 2013.

16. The eleventh mission was conducted by three members of the SC, Mr. Jean-Michel

Durr, Mr. Bent Noerby Bonde and Mr. Guido Pieraccini from 16 to 19 July 2013. The objec-

tive of the mission was to assess whether the preparations were still on track after the positive

assessment of the 10th mission to conduct a census in October 2013. A special focus was put

on the preparation of data processing.

17. The twelfth mission was conducted by four members of the SC, Mr. Roberto Bianchi-

ni, Mr. Bent Noerby Bonde, Ms. Katerina Kostadinova-Daskalovska and Mr. Guido Pieracci-

ni from 3 to 6 September 2013. The objective of the mission was to assess the state of play

regarding to forthcoming census field activities (including PES) and data processing activities

following the field activities as well as the overall census preparations in accordance to the

Census milestones and SC recommendations and advices given through the previous mis-

sions and SC reports.

18. The thirteenth mission was conducted by five members of the SC: Mr. Roberto

Bianchini and Mr. Guido Pieraccini from 27 September to 4 October, Ms. Katerina Kostadi-

nova-Daskalovska from 4 October to 11 October; and Mr. Jean-Michel Durr and Ms. Mer-

yem Demirci from 11 October to 18 October. The objective of the mission was to monitor the

enumeration of the census, including its final preparations and final operations. During this

mission, the SC was assisted by a team of 27 international observers, deployed all over the

country.

19. The fourteenth mission was conducted by one member of the SC: Ms. Katerina

Kostadinova-Daskalovska from 1 to 3 November 2013. The objective of the mission was to

monitor the conducting of the Post-enumeration survey following the Census. During this

mission, the SC was assisted by a team of 5 international observers, deployed all over the

country from 1 to 10 November.

20. The fifteenth mission was conducted by one member of the SC: Mr. Guido Pieraccini

from 17 to 18 December 2013. The objective of the mission was to assess that the data pro-

cessing system was functional and in accordance with the SC recommendations.

21. The sixteenth mission was conducted by five members of the SC from 11 to 14 Feb-

ruary 2014. The objective of the mission was to assess the starting of the data processing and

the implementation of the SC recommendations.

22. The seventeenth mission was conducted by one member of the SC: Mr. Guido Pierac-

cini from 21 to 23 May 2014. The objective of the mission was to control that the data pro-

cessing was working smoothly and that the timing for completion of the phase was in accord-

ance with the SC recommendations.

23. The eighteenth mission was conducted by four members of the SC from 3 to 6 June

2014. The objective of the mission was to assess the situation of data capture and the state of


preparation of further steps, editing and dissemination as well as the Post Enumeration Sur-

vey.

24. The nineteenth mission was conducted by two members of the SC, Ms. Katerina

Kostadinova-Daskalovska and Mr. Guido Pieraccini from 15 to 17 September 2014. The ob-

jective of the mission was to assess the data capture and editing phases, as well as PES and

dissemination preparation.

25. The twentieth mission was conducted by the complete SC from 18 to 21 November

2014. The objective of the mission was to assess in detail progress made in the data pro-

cessing phase, as well as PES and dissemination preparation.

26. The twenty-first and present mission was conducted by three members of the SC, Mr.

Jean-Michel Durr, Ms. Meryem Demirci and Mr. Guido Pieraccini from 24 to 27 February

2015. The objective of the mission was to assess in detail progress made in the data pro-

cessing phase, as well as PES and dissemination preparation.

27. The experts wish to express their gratitude to the directors and staff of the Agency for

Statistics of Bosnia and Herzegovina (BHAS), the Statistical Institutes of the Federation of

Bosnia and Herzegovina and Republika Srpska (RS) for their availability and collaboration,

and to the staff of Council of Europe for the organisation of the mission.

Executive Summary

28. Coding of activity, occupation and education, as well as geographic variables has

been performed since the last mission, with a good level of quality for geographical identifi-

ers and occupation and a reasonable one for industry. However, some issues remain to be

solved concerning coding of municipalities and settlements.

29. Coding of the three sensitive questions (ethnicity, religion and mother tongue) was

carried out in accordance with the recommendations of the SC to not proceed to any aggrega-

tion of answers other than gender or orthographic variants of the same word.

30. With the help of the two International Experts, the definition of the procedures of edit

and imputation to correct inconsistencies of the Census micro-data was carried out. Prelimi-

nary results in terms of frequency of edits activation’s and rates of imputation are satisfacto-

ry. Considering the delay already accumulated and the plan to publish the first census results

in May, the SC recommends to edit only sex and age in a first step and later the rest of the

variables in the database. Such an approach will probably allow to disseminate basic demo-

graphic data and the sensitive questions before summer 2015.

31. There was no significant progress in finalising the matching procedures between the

PES and the census data since the last SC mission mainly because of lack of coordination and

managerial decision. Actually, lots of time has been spent for matching operation because of

unavailability of decrypted census database which is needed for application of matching pro-

cedures to the whole census data. However, the SC appreciates good collaboration among the

PES team members and their professionalism in undertaking the activities.

32. There were not many developments in the preparation of the census data dissemina-

tion since the last SC mission. This is partly due to some activities that depend on the finali-

sation of other activities but mostly because of the lack of decisions regarding priority issues


such as procurement procedures for printing, the content of the first release and following

steps for the dissemination of census results including preparation of thematic reports.

33. The SC appreciates the high level of quality of the technical assistance and its effort

for providing efficient technical assistance in each mission.

34. Critical decisions remain to be taken, regarding the determination of the residence sta-

tus and the exclusion/inclusion in the database of some specific questionnaires with regard to

the Census law. The lack of management and decision making process hampers seriously the

finalisation of the editing process and thus the availability of a complete database ready for

tabulation and release of first results.

35. The SC reiterates its recommendation that at the end of the data processing the images

be destroyed and the personal data be removed from the Census database.

36. In conclusion, the SC considers that the data processing phase is at high risk of

serious delays and urges the Statistical Institutions to finalise decisions needed to com-

plete the editing phase in order to not further delay the dissemination of the census re-

sults.

Management and Decision Making Process

37. A new Director of the Agency has not been appointed yet and the two deputy direc-

tors are still acting Directors. Whereas the Central Census Bureau is meeting regularly, it is

not taking decisions on the critical issues.

38. The negative consequences on the overall management of the census and the decision

making process observed in the previous SC mission have worsened. Divergences among the

three SIs are still witnessed by the SC, in particular regarding the coding rules for the resi-

dence status and the rules related to access to the database. The data processing team has to

request specific authorisations to perform tabulations or editing tests on the complete data-

base, which causes delays in the preparations of further editing phases as well. Another con-

sequence of the inefficient management is that the Technical Assistance provided by the EU

is not used as efficiently as possible, as for example authorisation to decrypt the database for

matching unresolved cases of the Post Enumeration Survey was not given on time for the ex-

pert mission. Whereas the SC understands the need for strict control of access to the database

to prevent any leakage of information, this should be guaranteed by the oaths of confidentiali-

ty signed by the staff. However, despite this unfavourable managerial environment, the SC

could notice that staff of the three SIs is still motivated and dedicated.

Recommendations

39. The SC strongly recommend to fill the vacancy as Director of the BHAS as soon as

possible and reiterates it recommendation to establish a clear decision making process and to

give full authorisation to the data processing staff to perform its duty on the complete data-

base.


Data Processing

CODING

40. A global strategy was developed to perform the coding of the three main international

classifications adopted for the census (ISCO, NACE and ISCED) and an ad-hoc application

called g-Code was developed to approach these issues. The coding of Municipality, Settle-

ment, Country, Citizenship, Ethnicity, Religion and Mother tongue was also supposed to be

performed with the g-code application.

41. The coding operations of the geographic identifiers (country, municipality and settle-

ment) started in July 2015 using the g-code application but soon the three institutions realized

that the application was time consuming and decided to periodically export the data to be

coded in Excel and to hire additional coders controlled by their staff. The three Institutions

also realized that municipality should have been coded together with settlement due to fre-

quent mistakes of enumerators in filling the questionnaires (swap between municipality and

settlement or availability only of the settlement). So at the end, the coding operations of the

geographic identifiers took place separately from the database and in a semi-automated way.

42. Coding of the countries has been done in accordance with UN code list - Standard

Country or Area Codes for Statistical Use. All the records which have any connections with

Kosovo under UNSCR 1244/99 were coded with code 950. All the records that could not be

identified were coded with 999 (unknown).

43. Coding of the occupation was performed in accordance with the ISCO-08 classifica-

tion (four digits). The coding operations started in July with a thesaurus of 11.385 cases,

mainly coming from the last available LFS survey. At the end of the operations 70.1% of the

cases where automatically coded. The process of coding was completed by end of December

2015.

44. Coding of the industry has been done in accordance with NACE-rev 2.0 (two digits).

The coding operations started in July 2015 with a thesaurus mainly derived from the LFS

survey. The cases that were not automatically coded were first attempted to be coded on the

basis of the activity description and, if not possible, by searching for the company name in

the business register or in other administrative registers. The total amount of descriptions au-

tomatically coded was 454.872 (37.9%) over an amount of 1.199.525 cases.

45. Concerning the coding of the sensitive questions ethnicity, religion and mother tongue

the operations started in November and ended in February.

46. In its last report the SC reiterated that an assessment of the quality of the coding pro-

cess should have been performed verifying randomly 2-3 % of the cases of each variable

submitted to codifing.

Assessment

47. The assessment of the quality of the coding process for the geographic identifiers was

performed on different samples: i) 199.061 records corresponding to 6,4% of total records

having mention(s) of a country in the P1 database; ii) 325.912 records corresponding to 2,5%

of the total number of records having mention(s) of municipality in the P1 database; iii)


325.912 records corresponding to 2,5% of the total number of records having mention(s) of

settlement in the P1 database.

48. The results of the assessment of the quality for the eight questions in which was asked

the country is 0,3% errors out of the total number of records selected in the sample, with a

pick in question 19 (citizenship) of 0.6%.

49. The results of the assessment of the quality for the seven questions in which was

asked the municipality is 0,2% errors out of the total number of records selected in the sam-

ple, with a pick in question 13 (Place where the person resided during the 1991 census) of

0.4%.

50. The results of the assessment of the quality for the seven questions in which was

asked the settlement is 1,1% errors out of the total number of records selected in the sample,

with a pick in question 13 (Place where the person resided during the 1991 census) of 1.9%.

It should be noted that the above mentioned number of errors includes the cases in which the

municipalities was erroneously coded and, consequentially, the settlement was also errone-

ously coded. Therefore 1.1% could be seen as a quality indicator of both municipality and

settlement together.

51. Concerning municipality and settlement it was reported that at the time of the mission

a certain amount of cases were still not coded because of several ambiguous situations. The

amount of these cases is 660,525 over a total of 10,532,775 cases (6.3%), considering the

eight variables in which municipality and settlement were asked. At this stage the operations

of coding cannot be considered accomplished yet.

52. The assessment of the quality for the occupation was performed on a random sample

of 9% percent of different occupation (5% of the total records). The results show a total dif-

ferences of 10.3% of the cases of which 6.3% are the real differences while the remaining 4%

of the cases belongs to ambiguous descriptions that could be codified in different ways.

53. The assessment of the quality for the industry was performed on a random sample of

2% of each different activities (24.000). The results showed 3.658 (15.2%) wrong codes. It

should be noted that during the operation of validation the search in the business register for

the name of the company were not performed and this can have affected in certain way the

result.

54. The codification (meaning the operation of creating a classification based on the in-

formation collected) of the three sensitive questions was performed following the SC recom-

mendation to NOT proceed to any aggregation of answers other than gender or orthographic

variants of the same words. The observance of this recommendation was assessed by the SC

during the mission. A formal decision to adopt the codebooks was not taken by the Central

Census Bureau but the methodologists of the three institutions were in agreement and imple-

mented them in the database.

55. Ethnicity had originally 5.267 different entries (109.306 records) and at the end of the

codification process a total of 782 different codes remained in the codebook. Religion had

3.450 different entries (55.366 records) and at the end of the codification process a total 780

different codes remained in the codebook. Mother tongue had 3090 different entries (69.116

records) and at the end of the codification process a total of 764 different codes remained in

the codebook.


56. The SC assessed that the coding was performed with a good level of quality for geo-

graphical identifiers and occupation and a reasonable one for industry.

Recommendations

57. The SC recommends to solve urgently the remaining unresolved cases concerning the

codification of municipality and settlement so that a final raw database would be available as

soon as possible for the data editing phase.

DATA EDITING

58. The general objectives of the data editing strategy is to: i) verify and correct the inco-

herences between the answers related to P1 and P2; ii) verify and correct the incoherences in

the answers pertaining to each family nucleus and household; iii) identify the most probable

answers in case of missing values. Such corrections will be made by considering that some

incoherencies derive from systematic errors and that other errors should be corrected by con-

sidering a statistical method based on the principle of a minimum change in the values of the

variables and by finding the most probable value for the erroneous ones using a “donor ap-

proach”.

59. The first step of the editing strategy would be the determination of the resident status

for each individual and will be composed by the following sub-steps:

a. Application of deterministic corrections on Q1-Q7 variables to take into account some

incoherencies in the answers that can be solved with rules that imply a minimum

changes of variables;

b. Application of 29 different rules introduced to determine the resident status;

c. Use of a statistical method to correct the incoherencies on Q1-Q7 for all those indi-

viduals for which it was not possible to determine the Resident status;

d. New application of the 29 rules for the cases related to the step c;

e. Automatic assignment of a Not resident status for those individuals that, after step d,

still were not assigned a valid resident status.

60. The second step of the editing strategy would be the editing of all the variables in P2

and will be composed by the following sub-steps:

a. Editing and imputation considering all the qualitative variables for each households

b. Editing and imputation for each dwelling/building in order to obtain a coherence for

those variables involved in these checks, like number of floors, presence of the eleva-

tor, etc.

c. Editing and imputation for the quantitative variables referred to the characteristics of

the house (number of rooms, square meters, etc.)

f. Validation and corrections of the variables related to the agricultural section of the

questionnaire

61. The third and final step of the editing strategy would be the editing of all the variables

in P1, except Q1-Q7 and will be composed by the following steps:

a. Editing and imputation considering the relations among the records for a small subset

of variables: Age, Sex, Marital Status, and for those that identify the relationships be-

tween the households nuclei;


g. Editing and imputation considering the coherence of the variables inside the individu-

al records for all the variables, except Q1-Q7.

62. In certain cases probabilistic rules to correct an incoherent value may be changed to

deterministic rules as, for example, for the marital status that should be missing for each indi-

vidual less than 15 years old. Such deterministic rules can be derived also by observing the

questionnaires and how it was filled and should be applied before the probabilistic correc-

tions.

Assessment

63. The starting point for applying the data editing strategy will be the set of records that

will be considered valid according to the agreement between the three Statistical Institutions

on the conditions of exclusion/inclusion derived by the Census law. This means that the rec-

ords should be: i) without duplication; ii) validated according to the conditions of exclu-

sion/inclusion; iii) having a valid match between P1s and P2. At the moment there is still

presence of several not valid cases in the database (around 5.000) that need corrections and

manual interventions to be fixed. Moreover no formal decision is taken about the exclu-

sion/inclusion of cases derived from the Census law.

64. It should also be added that there is a pending decision on the criteria to be adopted

for identifying the usual resident population.

65. At the time of the mission no decision was yet taken regarding for which variables the

missing values will be kept and for which variables missing values will be imputed.

66. The results in terms of frequency of edits activation’s and rates of imputation for the

P2 records seem good. Moreover, the records detected without errors are about 65% against

35% with at least one error.

67. Some preliminary results in terms of frequency of edits activation’s and rates of impu-

tation for P1 records seem also to be good. The records detected without errors are about 53%

against 47% with at least one error. These results are to be considered preliminary since the

procedure for the correction of P1 is still under revision.

Recommendations

68. The presence of several not valid cases related to duplication and not valid match be-

tween P1s and P2 together with the non-agreement between the three Statistical Institution on

the conditions of exclusion/inclusion of some questionnaires in regard to the Census law is

preventing from finalizing the database. This situation risks producing serious delays in the

remaining steps of the census since it does not permit to calibrate properly the data editing

procedure and the SC recommends that urgent decisions be taken on these issues.

69. The non-agreement between the three Statistical Institution on the criteria to be adopt-

ed for determining the usual resident status is also risking to provoke serious delay. The SC

reiterates the recommendation of the 20th

report about the criteria for the determination of the

usual resident population.

70. Considering the delay already accumulated and the plan to publish the first census re-

sults in May, the opinion of the SC is that a two-step strategy for the data editing may be tak-

en in consideration which includes edit only sex and age in a first step and at later step the


rest of the variables in the database. Such an approach will probably allow disseminatation of

basic demographic data and the sensitive questions before summer.

71. The edit and imputation procedure should pass substantial tests before being approved

for use, including tests on the completed census materials. Every change of the census data

performed through the procedure of edit and imputation must be tracked into the Census da-

tabase and should be reversible. The SC should be able to assess, and eventually re-run on the

initial Census database the procedure of edit and imputation defined for the correction of log-

ical errors.

72. The SC recommends that a full assessment of the impact of the data cleaning proce-

dure will be implemented as soon as the procedure will be applied to the complete set of data.

The assessment should be based on the comparison of the initial set of data with the set of

coherent, cleaned data obtained applying the whole edit and imputation procedure. Indicators

at aggregate level, which take into account the changes produced in terms of number and/or

magnitude, should be adopted. Indicators at variables level to underline the differences pro-

duced in the distributions of each variable should also be considered.

Personal information and census materials

73. The request made after the 16th

mission in February 2014 to the BHAS to analyse the

legal framework related to archiving questionnaires and present to the SC a proposal of

treatment of personal data (images, paper questionnaires and database) was not yet fulfilled.

Recommendations

74. The SC strongly reiterates its recommendation to the Agency to analyse the legal

framework related to archiving questionnaires and files and to present to the SC a proposal of

treatment of personal data (images, paper questionnaires and database).

75. The SC reiterates its recommendation that at the end of the data processing the images

be destroyed and the personal data be removed from the Census database. Indeed, since the

paper questionnaires will be archived, there is no need to maintain any digital version of them

for historical reasons. If needed, the SC considers it would be worth amending the census law

to ensure the strict protection of individual data and protect the Statistical Institutions from

any suspicion of misuse of the data collected.

Post Enumeration Survey

76. Matching procedures and steps developed by the PES team for matching the PES and

the census data were validated through application of the software RELAIS, an open-source

toolkit for record linkage offering consolidated statistical procedure, with assistance of the

TA. Comparison of the already achieved matching results with the outcomes of the software

have indicated very high level of linkage results with overlapping matching records of 99.6

percent.

77. According to the agreed methodology, first step of matching of individuals, house-

holds and dwellings enumerated in the PES with the PES EAs of census data as well as with a

set of neighboring EAs would be followed by matching the unmatched individual in PES

against the whole census database. For the implementation of this phase, as the PES expert in

the TA teamindicated in the first mission report, the whole census data has to be provided as


decrypted in order to be matched with PES data. However, the decrypted data was not ready

during the expert`s mission which was conducted in 16-20 February, therefore it was not pos-

sible to start this activity as planned. The decrypted database will be available by the end of

March 2016.

78. As suggested by the expert, the PES team analysed the unmatched individuals in the

corresponding EAs of the census database to investigate reasons for the high amount of un-

matched records on the basis of characteristics of individuals and households. According to

preliminary results of this analysis, as it could be expected, there is high correlation between

unmatched individuals and two indicators, namely size of household and average age of

household. The PES team will continue this analysis to understand the behaviour of house-

holds over-enumerated in the census.

79. The PES team made simulations for the application of the 29 rules identified by the

methodologist group for identification of usual residence status using the questions 1 to 7 in

the individual census questionnaire to analyse to what extent the application of usual resi-

dence status would affect proportion of unmatched cases particularly in the PES EAs of the

census database.

80. There is a plan for disseminating main results of coverage errors together with the

first results of the census. According to current plan, measurements of content errors will be

disseminated in the quality assessment report which will be prepared in the context of themat-

ic reports. The PES team emphasizes that they do not have experience in estimating content

errors; therefore they do not have any idea about how long this task may take.

81. The PES team expressed the need for reviewing the technical assistance work plan in

order to increase the allocated time for PES activities in particular for the calculation of

weights, estimation of coverage and content errors and analysis of results. The TA will re-

view current plan and make necessary adjustment in the project in order to provide adequate

support.

Assessment

82. There was no significant progress in finalising the matching procedures since the last

SC mission mainly because of lack of coordination and managerial decision. Actually, lots of

time has been spent for matching operation because of unavailability of decrypted census da-

tabase needed for application of matching procedures to whole census data. In addition, inef-

fective decision making mechanism regarding the finalisation of the rules for determining

usual residence status and for inclusion/exclusion of some population groups have signifi-

cantly delayed the implementation of PES activities.

83. The SC appreciates good collaboration among the PES team members and their pro-

fessionalism in undertaking the activities. The SC also appreciates the support of the TA and

its effort for providing efficient technical assistance in each mission by giving pre-conditions

that should be fulfilled before conducting a mission.

Recommendations

84. The SC recommends that the request of the PES team for extending the allocated time

for the remaining PES activities be considered by the TA and necessary adjustment in the

project be done timely.


85. The coordinators of the census project should carefully follow up the pre-conditions

given by the TA and make sure that these conditions are achieved timely; otherwise the TA

missions should be postponed until the requirements for conducting a mission are fulfilled.

Dissemination of census results

86. The inter-institutional dissemination working group reviewed the plan for thematic

reports with the support of the TA. The working group agrees to reduce the number of re-

ports from 16 to 9 reports by combining some of related census topics in one publication.

They are also reviewing the content of the reports for presenting mainly analytical results

based on indicators and summary statistics.

87. The tender for printing census reports and advertisement materials is still not in pro-

cess. The dissemination working group is waiting for final decisions for number of publica-

tions and pages to determine specifications for the procurement.

88. The tabulation plan for consistency control of census data was prepared with the sup-

port of the TA. However, the experts on dissemination were not able to access the tables and

required authorization for accessing the census results in table format.

89. The Census Central Commission agreed on releasing main results of the census with 9

tables at its meeting on February 11, 2015. The dissemination working group also considers

the possibility of expanding the number of tables up to 26 for the first release of census re-

sults according to the suggestion of the TA. This situation gives the impression that the tabu-

lation plan for the first release of census results is still under discussion.

90. The first census results will be released in a press conference with participation of all

partners, media and main users of the census results.

Assessment

91. There were not many developments in the preparation of the census data dissemina-

tion since the last SC mission. This is partly due to some activities that depend on the finali-

sation of other activities but mostly because of the lack of decisions regarding priority issues

such as procurement procedures for printing, the content of the first release and following

steps for the dissemination of census results including preparation of thematic reports.

Recommendations

92. The SC reiterates all recommendations and advices given in the last SC report refer-

ring to dissemination and its advices related to the plan for preparation of metadata in order to

enable quick preparation of online dissemination and publications, once the final census data

is available.

93. The SC strongly suggests that clear description of procedures for storing and access-

ing the tables produced for validation and dissemination be prepared. Authorisation for ac-

cessing these tables should be given to TA for assisting consistency control and to the SC for

assessing the census results.

94. The SC supports the idea of releasing the first results in a press conference instead of

a seminar if the first results cover few census topics.


ANNEXE

AGENDA OF THE MISSION

International Monitoring Operation Steering Committee

Date: 24 - 27 February 2015

Tuesday, February 24, 2015, Data Processing Centre, East Sarajevo

09:00 – 13:00 Meeting with the coordinators of Data Processing

Centre Data processing (current status)

BHAS: Admir Kulic, Nora Selimovic

FIS: Hidaeta Colovic, Jasmina Velic

RSIS: Bogdana Radic, Rajko Vranjes

Data Processing

Centre, East

Sarajevo

09:45 – 10:45 Meeting with EU Delegation

Jean-Michel Durr

13:00 – 14:00 Lunch break

14:00 – 17:00 Meeting with Methodology and IT Team Data coding, editing and control of some of records

BHAS: Vladimir Mijovic, Radoslav Corovic, Anel

Grabovica (methodology); Sevala Abaz, Vesna Sehovac

FIS: Semso Surkovic, Nermina Sarajkic, Dzemal

Borovina, Asmira Selimovic, Muamera Idrizovic, Emina

Ajkunic

RSIS: Dara Petkovic, Rajko Vranjes, Bojan Stankovic,

Radoslav Latincic

Data Processing

Centre, East

Sarajevo

Wednesday, February 25, 2015, Data Processing Centre, East Sarajevo

9:00 – 12:00 Meeting with Methodology and IT Team (continuation)

Data coding, editing and control of some of records

BHAS: Radoslav Corovic, Vladimir Mijovic, Anel




Ajkunic


Radoslav Latincic

Data Processing

Centre, East

Sarajevo

14:00 – 17:00

Meeting with Methodology and IT Team (continuation)

Data coding, editing and control of some of records

BHAS: Radoslav Corovic, Vladimir Mijovic, Anel




Data Processing

Centre, East

Sarajevo


Ajkunic


Radoslav Latincic

Thursday, February 26, 2015, Agency for Statistics of Bosnia and Herzegovina

9:00–12:30 Meeting with dissemination and communication

team: Preparations and plans for dissemination;

BHAS: Ivona Bulic; Alen Mrgud

FIS: Zehra Kaba, Mevla Softic, Dinko Demirovic, Amer

Koric

RSIS: Vladan Sebinovic, Dragana Mandic, Igor Radan

BHAS, Sarajevo

12:30 – 14:00 Lunch

14:00-16:00 Meeting with dissemination and communication (cont)

Seminar of Presentation of first results (in April or

May?): preparation, agenda (continuation)

BHAS: Ivona Bulic; Alen Mrgud

FIS: Zehra Kaba, Mevla Softic, Dinko Demirovic, Amer

Koric

RSIS: Vladan Sebinovic, Dragana Mandic, Igor Radan

BHAS, Sarajevo

Friday, February 27, 2015

9:00 – 12:00 Meeting

Meeting with PES Team

Status, estimates…

BHAS: Edin Sabanovic, Jasna Isakovic, Marko Zekic

FIS: Arijana Amina Ramic, Muamera Idrizovic, Amel

Sikiric

RSIS: Darko Marinkovic, Dean Arezina

BHAS, Sarajevo

12:00 – 13:00 LUNCH

13:00 – 16:00 Meeting with the directors of the three statistical

institutions

(BHAS, FIS, RSIS)

BHAS: Fadil Fatic

FIS: Hidajeta Bajramovic, Galiba Karacic

RSIS: Radmila Cickovic, Radosav Savanovic

BHAS, Sarajevo

steering committee of the international monitoring ...bhas.ba/census/1348896-report 21 - sc mission...

Documents