Almiro Moreiraalmiro.moreira@ine.ptStatistics Portugal

«

25 September 2013

Geneva, Switzerland

UNITED NATIONSECONOMIC COMMISSION FOR EUROPE

CONFERENCE OF EUROPEAN STATISTICIANS

Topic (ii): Centralising data collection

CREATING A DATA COLLECTION DEPARTMENT:

STATISTICS PORTUGAL'S EXPERIENCE

Paulo Saraiva dos Santos

paulo.saraiva@ine.ptStatistics Portugal

Sharing eight years of experience in centralising data collection,

implementing an integrated and process driven approach to

change the statistical production, improving its efficiency and

flexibility.

Motivation

2

1. Background and context;2. Reengineering the production;3. Centralised Data Collection;4. Administrative Sources &

registers;5. Data Collection infrastructure6. Benefits of a Centralised

Approach;7. The future of Data Collection.

Outline

3

• Central national authority for the production of official statistics;

• Aims at developing and supervising the national statistical system;

• Created in 1935, has its head office in Lisbon with delegations in Porto, Coimbra, Évora and Faro.

Statistics Portugal (INE)

4

Madeira

Azores

Oporto

Lisboa

Coimbra

Évora

Faro

Geographical dispersion

5

Common technical

requirements, methods and infrastructure

• Statistics Portugal is a public institution which has legal personality, administrative autonomy and technical independence in the exercise of its official statistical activity;

• It is a special public institution integrated within indirect State administration;

• The Statistical Law confers on Statistics Portugal statistical authority and legal obligation to confidentiality.

Statistics Portugal

6

• European and National Statistics (2013):

European scope

7

Statistical Operations (2013) European National / Other Total

Statistics Portugal 153 82% 33 18% 186 100%

82%

18% Quantity of Statistical Operations

Timeline

8

• From 1989 to 2003: – Headquarters & Regional Directorates;

• Regional Directorates:– Firstly acting as dissemination and data collection center

for the region (NUTS II);– Gradually assumed active role in statistical production

and regional studies;– Its organization and resources have increased fast.

• From 2003:– Proactive evaluation of the existing model;– Reorganization not guided by resources

constraints.

Reorganization in 2004

9

• New Executive board in 2003;• Hired external advisory company

(international strategy consultants);• Request a Peer Review in 2004:

– Mr. Ivan Fellegi • Former Chief Statistician of Canada from 1985 to

2008;– Mr. Jacob Ryten

• Former Assistant Chief Statistician of Canada from 1969 to 1997.

• A proactive action: to create a new structure.

Former organization (2003)

10

Executive Board

Lisbon and Tagus Valle North

Center Alentejo

Algarve

Dissemination

Finance Human Resources

Planning and International

Legal Support Methodology

Information Systems

National Accounts Agriculture

Population and Census Business

Industry and Services Social

Short Term and Forecast

Regional Directorates Support Subject Matter

Regional Directorates (2003)

11

Regional Directorat

e

Social Business Studies

RH & ResourcesIT Support

Dissemination

Regional Directorate (Department)Unit

Section

Three hierarchical levels

An example of the organization of a former RD.

Local 1

Survey 2 Survey n

12

Difusão

Tratamento

Recolha

Difusão

Tratamento

Recolha

Difusão

Tratamento

Recolha

...

...

Former architecture

Difusão

Tratamento

Recolha

Difusão

Tratamento

Recolha

Difusão

Tratamento

Recolha

Dissemination

Treatment

Collection

Dissemination

Treatment

Collection

Dissemination

Treatment

Collection

Local 2

Local n

Survey 1

...

Stovepipe systems

Complex, inefficient and not flexible

Former organization (2003)

13

• Heavy and costly organization– 788 workers: 37% in Regional Directorates.– 195 managers (25%):

• 14 Departments, 5 Regional Directorates, 48 Units, 128 sections • Duplication of work, procedures and tools;• Not flexible enough for the future.

Need to be reorganized

Fellegi & Ryten’s Peer Review

• Objective: to review the Portuguese statistical system and produce recommendations;

• Main results:• The diagnosis;• Structural problems and

remedies;• Recommendations

• Started in 2004 and based on the Peer Review´s recommendations;• Internal reorganization:

– A central data collection department was created;– Regional directorates were extinct;– Domain departments have been merged into three units: economics, social and national accounts;– Methods and information system were merged into one department.

• It was a successful challenge, although some resistances and constraints.

15

Production re-engineering

New organization (2004 2013)

16

Executive Board

Porto

Finance & HR

Inf Systems Methodology

Data Collection

National Accounts

Economics

Social

Delegations Support Statistical Production

CoimbraÉvora

Faro Subject Matter

Staff

Dissemination

PlanningLegal

SupportInternationalCommunicati

on

L1: DepartmentL2: UnitL3:

Section

Three hierarchical levels

Production architecture

17

Socia

l and

De

mog

raph

ics

Econ

omics

Natio

nal

Acco

unts

Data Collection

Met

hods

and

In

form

atio

n Sy

stem

s

18

Impact of the reorganization

2003 2013 % Diference

L1: Departments 19 7 -63%

L2: Units 48 34 -29%

L3: Sections 128 13 -90%

Managers 195 61 -69%

Workers / Managers 4,0 10,9 173%

Workers 788 665 -16%

Lisbon 496 508 2%

Regions 292 157 -46%

Staff reduction without firing anyone

19

Human Resources Distribution(by macro process)

Production58,9%

Support31,9%

Staff9,2%

Centralised Data Collection

• Survey’s data collection:– 40% budget & 30% human resources.

Data Collection at Statistics Portugal

21

Survey Data Collection is a core function

• A Data Collection department assures the collection, processing and analysis of collected microdata, covering all business and social surveys;

HR ~ 200 workers + 350 freelance interviewers

Data Collection

22

120 surveys 105 business (self-completed) 15 by interview (CAPI and CATI).

125.000 companies (99% SME); 70.000 dwellings; 35.000 farms.

Ann

ual f

igur

es

Data Collection Department

23

Data Collection

Self-completed

Surveys

Interview Surveys

Data CollectionProcesses

Lisbon 1

Lisbon 4

Lisbon 3

Lisbon 5

Lisbon 6

Lisbon 7

CoimbraPorto 1

Porto 3

Évora Faro

Porto 2

Data Collection DepartmentHuman Resources by Unit

24

Self-completed surveys

46%

Staff1.5%

Interview Surveys

40%

Data Collection Processes13%

Data Collection DepartmentOrganization by Unit

25

• Self-completed surveys:– By project or statistical operation;– National management of each project;

• Interview surveys:– Sections work with the same projects;– Share same methods, procedures and

tools.• Data collection processes;

– National coordination of interview surveys;

– CATI national coordination.

Management within DC

26

• Decentralized managed but centrally controlled;

• One overall budget distributed through each management level;

• Autonomy with responsibility;• Objective definition in “cascade”;

– Department Unit Section worker• HR: matrix management;

Interview Management System

27

• Interview Management System supports all the processes related with social statistics and the price collection;

• The Survey Management System has several components: team management and the tools used by the interviewers to collect data, transfer them to Statistics Portugal, allowing them to work both in face-to-face and telephone interviews.

• .

HR and costs control

28

• Assiduity control WebRH app;• Accounting to projects Factiv app

– Project codes and Task codes;– Individually daily allocation of the working time

to each project code and tasks;• Direct HR costs are monthly calculated to

each project, according to individual wages and social costs;

• The same with other costs and indirect costs;

• Transfers can be made between projects.

2 Design

3 Build

4 Collect

5 Process

6 Analyse

7 Disseminate

1 Specify

Needs

3.5 Test statistical business process

3.4 Test production

system

3.3 Configure workflows

3.2 Build or enhance process

components

3.1 Build data collection instrument

1.6 Prepare business

case

1.5 Check data

availability

1.3 Establish

output objectives

1.2 Consult and confirm

needs

1.1 Determine needs for

information

2.6 Design production

systems and workflow

2.5 Design statistical processing

methodology

2.4 Design frame and

sample methodology

2.3 Design data collection

methodology

2.2 Design variable

descriptions

2.1 Design outputs

4.4 Finalize collection

4.3 Run collection

4.2 Set up collection

4.1 Select sample

5.1 Integrate data

5.2 Classify and code

5.3 Review, validate, edit and analyze microdata

5.4 Impute

5.5 Derive new variables and statistical

units

5.6 Calculate weights

5.7 Calculate aggregates

6.1 Prepare draft outputs

6.2 Validate outputs

6.3 Scrutinize

and explain

6.4 Apply disclosure

control

6.5 Finalize outputs

7.5 Manage user support

7.4 Promote dissemination

products

7.3 Manage release of

dissemination products

7.2 Produce dissemination

products

7.1 Update

output systems

8 Archive

9 Evaluate

8.2 Manage archive

repository

8.1 Define archive rules

8.3 Preserve data and associate metadata

8.4 Dispose of data and associated metadata

9.1 Gather evaluation

inputs

9.2 Conduct evaluation

9.3 Agree action plan

Levels 1 and 2GSBPM, version 4.0

1.4 Identify concepts

3.6 Finalize production

system

5.8 Finalize data files

Data Collection

Subject Matter

Shared DC/SM

IS & Methods

Dissemination

Quality Control

Division of work

Relationship between DC & Subject Matter (1)

30

• One major issue at the beginning;• There were a negative perception of

the DC tasks “a low profile work …”• Conversely, subject matter

statisticians were very “data collection oriented”;

• But expectations are always high!– “You have to do better than me (when I

was responsible for DC) …”

Relationship between DC & Subject Matter (2)

31

Solution• Service Level Agreements (SLA) to

manage expectations and to build trust;

• It was used a step-by-step approach, from a simplified version and increasing gradually the complexity.

Administrative Sources & Registers

Administrative Sources (ADS)

33

• ADS are not (still) in the scope of the DC Department;

• It is managed by Subject Matter departments, supported by IS & Methods;

• Statistics Portugal is still very “survey oriented”. Thus, ADS are not well developed;

• But there one remarkable initiative:– IES: Simplified Business Information

Data CollectionInfrastructure

• Survey Management System (SIGINQ);

• Other Data Collection Systems:– Datawahouse;– HomeCATI;– Interview Management System;– Telephone Data Entry;

Outline

35

Survey Management System

Survey Management

• Design a new approach of production based on a broad integration with process and tools standardization;• Use of an internal reference model to describe the statistical business processes (SPPM);

37

Re-engineering Working Group

Survey Management

Business Agriculture Social

• Management and control of all data collection processes, including information about respondents and paradata;

• Supported by the Metadata System;

Process Management System

Business Agriculture Social

SAGR

• Similar features, but adapted by statistical unit.

38

• GPap is the core for Business Surveys, linked with:– Questionnaires and Capture (WebInq and WebReg);– Respondent Management (GRESP), – Business Register (FUE),

• Transfers validated microdata to Datawarehouse.

Process Management System

39

Survey Unit Occur Collec

tRepor

tAnalysi

sUpdat

eMana

g. Help

Errors

Status

Validations

Primary Val

Upload

Insert

Manage entries

Data Entry

Method

Specific

Generic

SIGUA block prop

Manage

Cross

Specific

Supplement

Launch

By mode

Consult

Open / Close

Specific Tables

Generic Tables

Specific Reports

Generic Reports

Consult transfers

Transfer to analysis

Consult Analysis

GPap

Survey

Register

Sample

Respondent

Batch update

Table Manag.

Common process

Specific process

GPap components

40

41

BEA FNA

Survey Management

Contact Centre

Business Agriculture Social

SAGRInterviewer

Management

HomeCATI

42

• HomeCATI is an infrastructure which allows freelance interviewers work at home, integrated in a virtual contact centre and based on a voice over Internet protocol (VoIP) solution;

• This solution has many advantages, but there are many challenges to deal with, like the interview supervision and monitoring.

Interview Management System

43

• Interview Management System supports all the processes related with social statistics and the price collection;

• The Survey Management System has several components: team management and the tools used by the interviewers to collect data, transfer them to Statistics Portugal, allowing them to work both in face-to-face and telephone interviews.

• .

Telephone Data Entry (1)

44

• Telephone Data Entry (TDE), which is a solution by which respondents can return their data using the keypad on their telephone;

• Respondents are sent a letter which informs them of the free phone telephone number to call, their unique respondent identification key number, and the data required. On calling the telephone number, the respondent can choose the appropriated survey, and a recording of the survey questions is heard and the respondent enters their data using their telephone keypad.

Automated Data Collection

45

• INE is developing and implementing Automated Data Collection Methods for Business Surveys;

• It aims to reduce the reporting burden businesses, to improve the timeliness and to promote a more efficient way of collection data;

• Based on XML, it is already available for two surveys.

Benefits of Centralised Data

Collection

1. Development and management of a common infrastructure, both intellectual and operational, which could only be duplicated geographically;

2. Creation of a flexible, dynamic and responsive production architecture tied to the common services provided by shared means of production (sampling frames, classifications and standards, questionnaire designs, methods and tools, etc.);

47

Benefits of centralised DC (1)

3. Creation of right means of coordination to make our design work in order to face future (but now present) budgetary cuts;

4. Adoption of a cost-effective approach that makes the most effective use of regional and central resources. It was possible to do more with the same.

5. Reduction of the data collection cycle, specially the time to deliver statistical results;

48

Benefits of centralised DC (2)

6. Assistance to develop a steady culture based on efficiency and innovation, considering the full in-house design and development approach;

7. Development of analytic competences in order to improve the quality of the information (more reviewing and validation tools);

49

Benefits of centralised DC (2)

8. Creation of an integrated Survey Management System as well as other Data Collection tools;

9. Reduction of respondent burden:– Avoiding duplication of variables and offering easy

and multiple ways to provide data;10.Reduction of production costs;

– Estimated in 27.2% (business surveys; 2005 – 2012).

50

Benefits of centralised DC (2)

Cost reductionBusiness data collection

51

Total costsBase 2005 = 100%

- 27.2%

2005 2006 2007 2008 2009 2010 2011 20120%

25%

50%

75%

100%Chart Title

Electronic Data Collection

52

Visits

% electronic collection

Questionnairs

2013 – 100%

Electronic Data CollectionAvoid variable duplication

53

Common

Variables

Easy update

Future of Data Collection

• Increase the use of administrative sources;• Extend Integrated Production Systems;• Improve Automated Data Collection and the use of

Scanner Data on price collection;• Increase the multimodal collection capability (web

based);• Improve the use of paradata to support the quality

processes;• Create new processes to better understand respondent's

behavior in order to motivate their collaboration.

55

Future of Data Collection

Almiro Moreiraalmiro.moreira@ine.pt

Statistics Portugal

«

25 September 2013

Geneva, Switzerland

UNITED NATIONSECONOMIC COMMISSION FOR EUROPE

CONFERENCE OF EUROPEAN STATISTICIANS

Topic (ii): Centralising data collection

CREATING A DATA COLLECTION DEPARTMENT:

STATISTICS PORTUGAL'S EXPERIENCE

Paulo Saraiva dos Santospaulo.saraiva@ine.ptStatistics Portugal

Thank you for your attention!