Transcript Data Documentation Initiative: A global phenomenon

Data Documentation Initiative: A
global phenomenon – coming
soon to ABS!
Wendy Thomas
Chair, DDI Technical Implementation
Committee
1 September 2009
Acknowledgments
• Slides provided for use by:
– Wendy Thomas
– Pascal Heus
– Mary Vardigan
– Peter Granda
– Nancy McGovern
– Jeremy Iverson
– Dan Smith
What is Metadata?
• Common definition: Data about Data
Unlabeled stuff
Labeled stuff
The bean example is taken from: A Manager’s
Introduction to Adobe eXtensible Metadata Platform,
http://www.adobe.com/products/xmp/pdfs/whitepaper.pdf
Managing data and metadata is challenging!
We are in charge of the
We
want
access
need
to
collect
the
data.
We easy
support
our
to
high
quality
and
well
information
from
the
Webut
have
an
users
also
need
to
documented
data! it,
producers,
protect
our preserve
information
and
provide access to
respondents!
management
our
users!
Academic
problem
Producers
Users
Government
Sponsors
Librarians
Business
Policy Makers
General Public
Media/Press
Summary of ABS Metadata
Management Principles
• Life-cycle focus
• Data supported by
accessible metadata
• Metadata available and
useable in context of
client’s need
• Registration authority for
metadata element
• Clear identification,
ownership, approval
status of metadata
elements
•
•
•
•
Describe metadata flow
Reuse metadata
Capture at source
Capture derivable
metadata automatically
• Ensure cost/benefit of
metadata
• Variations from standards
tightly documented
• Make metadata active to
the greatest possible
extent
NISO: A FRAMEWORK OF GUIDANCE FOR
BUILDING GOOD DIGITAL COLLECTIONS
Metadata Principle 1: Good metadata conforms to community standards
in a way that is appropriate to the materials in the collection, users of the
collection, and current and potential future uses of the collection.
Metadata Principle 2: Good metadata supports interoperability.
Metadata Principle 3: Good metadata uses authority control and content
standards to describe objects and collocate related objects.
Metadata Principle 4: Good metadata includes a clear statement of the
conditions and terms of use for the digital object.
Metadata Principle 5: Good metadata supports the long-term curation and
preservation of objects in collections.
Metadata Principle 6: Good metadata records are objects themselves and
therefore should have the qualities of good objects, including authority,
authenticity, archivability, persistence, and unique identification.
Some major XML metadata specifications
for data content management
• Statistical Data and Metadata Exchange (SDMX)
– Macrodata, time series, indicators, registries
– http://www.sdmx.org
• Data Documentation Initiative (DDI)
– Microdata (surveys, studies), aggregate, administrative data
– http://www.ddialliance.org
• ISO/IEC 11179
– Semantic modeling, concepts, registries
– http://metadata-standards.org/11179/
• ISO 19115
– Geography
– http://www.isotc211.org/
• Dublin Core
– General resources (documentation, images, multimedia)
– http://www.dublincore.org
Metadata provides support for:
•
•
•
•
•
•
•
Survey and data collection preparation
Data collection
Data processing
Analysis
Data discovery and access
Replication
Repurposing (secondary data use or data
products)
Metadata
• Metadata is essential information for
research and reuse of data
• The further data gets from its source, the
greater the importance of the metadata
• Content is critical
• Structure is becoming increasingly
important in a networked world
Why Standards?
• Standards provide structure for:
– Accurate transfer of content between systems
– Increased automation of ingest, reducing
costs
– Interoperability between systems and
software
– Structural base for discovery and comparison
Example: Dublin Core
• Print card catalogs
• Standalone
databases
• WorldCat and Google
•
•
•
•
Static
stationary
Proprietary structure
Little cross-site
searching
• Standardized content
• Cross-site searching
Interacting Standards for Data
• Dublin Core
• ISO/IEC 11179
• ISO 19115 –
Geography
• Statistical Packages
• METS
• PREMIS
• SDMX
• DDI
• Citation structure
• Coverage
– Temporal
– Topical
– Spatial
• Location specific
information
Interacting Standards for Data
• Dublin Core
• ISO/IEC 11179
• ISO 19115 –
Geography
• Statistical Packages
• METS
• PREMIS
• SDMX
• DDI
• Structure and content
of a data element as
the building block of
information
• Supports registry
functions
• Provides
– Object
– Property
– Representation
Interacting Standards for Data
• Dublin Core
• ISO/IEC 11179
• ISO 19115 –
Geography
• Statistical Packages
• METS
• PREMIS
• SDMX
• DDI
• i.e., ANZLIC and US
FGDC
• Focus is on
describing spatial
objects and their
attributes
Interacting Standards for Data
• Dublin Core
• ISO/IEC 11179
• ISO 19115 –
Geography
• Statistical Packages
• METS
• PREMIS
• SDMX
• DDI
• Proprietary standards
• Content is generally
limited to:
– Variable name
– Variable label
– Data type and
structure
– Category labels
• Translation tools used
to transport content
Interacting Standards for Data
• Dublin Core
• ISO/IEC 11179
• ISO 19115 –
Geography
• Statistical Packages
• METS
• PREMIS
• SDMX
• DDI
• Digital Library
Federation
• Consistent outer
wrapper for digital
objects of all type
• Contains a profile
providing the structural
information for the
contained object
Interacting Standards for Data
• Dublin Core
• ISO/IEC 11179
• ISO 19115 –
Geography
• Statistical Packages
• METS
• PREMIS
• SDMX
• DDI
• Preservation
information for digital
objects
Interacting Standards for Data
• Dublin Core
• ISO/IEC 11179
• ISO 19115 –
Geography
• Statistical Packages
• METS
• PREMIS
• SDMX
• DDI
• Developed for
statistical tables
• Supports well
structured, well
defined data,
particularly timeseries data
• Contains both
metadata and data
• Supports transfer of
data between
systems
Interacting Standards for Data
• Dublin Core
• ISO/IEC 11179
• ISO 19115 –
Geography
• Statistical Packages
• METS
• PREMIS
• SDMX
• DDI
• Version 3.0 covers
life-cycle of data and
metadata
• Data collection
• Processing
• Management
• Reuse or repurposing
• Support for registries
• Grouping &
Comparison
Metadata Coverage
•
Dublin Core
•
ISO/IEC 11179
•
ISO 19115
•
Statistical Packages
•
METS
•
PREMIS
•
SDMX
•
DDI
•
•
•
•
•
•
[Packaging]
Citation
Geographic Coverage
Temporal Coverage
Topical Coverage
Structure information
– Physical storage description
– Variable (name, label, categories, format)
•
•
•
•
•
•
•
Source information
Methodology
Detailed description of data
Processing
Relationships
Life-cycle events
Management information
The Data Documentation
Initiative (DDI)
• International XML based specification
– Started in 1995, now driven by DDI Alliance (30+
members)
– Became XML specification in 2000 (v1.0)
– Current version is 2.1 with focus on archiving
(codebook)
• New Version 3.0 (2008)
– Focus on entire survey “Life Cycle”
– Provide comprehensive metadata on the entire
survey process and usage
– Aligned on other metadata standards (DC, MARC,
ISO/IEC 11179, SDMX, …)
– Include machine actionable elements to facilitate
processing, discovery and analysis
Intent of DDI Design
• Facilitate point-of-origin capture of metadata
• Reuse of metadata to support:
– Consistency and accuracy of metadata content
– Provide internal and external implicit comparisons
– Support external registries of concepts, questions,
variables, etc.
– Metadata driven processing
• Provide clear paths of interaction with other
major standards
Basic Structures
• DDI 3 used a model similar to SDMX in
terms of the following:
– Indentifiable, Versionable, and Maintainable
objects
– The use of multiple schemas to describe
different process sub-sections in the life-cycle
– Use of schemes to facilitate reuse of common
materials
DDI: Full content coverage for
survey and administrative data
•
•
•
•
•
•
•
•
•
Conceptual coverage
Methodology
Data Collection
Processing – cleaning, paradata
Recoding and derivations
Variable and tabular content
Internal relationships
Physical storage
Data management
Plus: Relationships between
studies
• Comparison by design
– Study series can inherit from earlier metadata
– Capture changes only
• Data integration
– Mapping of codes between source and target
– Capture comparison information
• Comparison of abstract content models
– Publication of reusable materials (code
schemes, concept schemes, geographic
structure, etc.)
Current Areas of DDI Development
• Controlled vocabularies to improve
machine actionability
• Data collection methodology and process
expansion for more depth and detail
• Qualitative data
• Increased comparison coverage
• Tools
DDI 3.0 Metadata Life Cycle
• Data and metadata creation is not a static process: It dynamically
evolved across time and involves many agencies/individuals
• DDI 2.x is about archiving, DDI 3.0 focuses on the entire “life cycle”
• 3.0 emphasizes metadata reuse to minimize redundancy and
discrepancies, support comparison, and drive the data and metadata
creation process
• Supports multilingual, grouping, geography, and registries
• 3.0 is extensible
When to capture metadata?
• Metadata must be captured at the time the event
occurs! (not after the facts)
• Documenting after the facts leads to
considerable loss of information
• This is true for producers and researchers
Reuse
• DDI is designed around schemes (lists of
items) for commonly reused information
within a study such as categories, code
schemes, concepts, universe, etc.
– Items are “used” in multiple locations in a DDI
document by referencing the item in the list
– Enter once, use in multiple locations
– Items can be versioned for management over
time without having to change content in
multiple locations
Comparison and Registries
• Information in DDI schemes can be
published in external registries and used
by multiple studies
– Provides implicit comparison both within a
study and between studies
– Supports organizational consistency through
the use of agreed content managed in
registries
– Referencing structured lists provides further
context to individual items used in a study
Metadata driven processing
• Capturing metadata upstream can provide
over 90% of the building blocks needed to
generate the remainder of the metadata
• DDI supports imbedding command code to
run data processing events driving data
capture, data processing during after
collection, and to support post-collection
recoding, derivations, and harmonization
maps
Questions to Variables
REGISTRY
Question
Development
Software
Instrument
Development
Software CAI
Identifying
Universe and
Concepts
Organizing
questions and
flow logic
Building or
Importing
Question Text
and Response
Domains
DDI
Capturing raw
response data
and process
data
Data Processing
Software
Data cleaning
and verification
DDI
Recoding and/or
deriving new
data elements
using existing or
new categories
or coding schemes
Working with other standards
• There is no single standard that does it all
• DDI was specifically designed to support
easy interaction with:
– Dublin Core – mapping of citation elements
and imbedding native Dublin Core
– ISO/IEC 11179 – working with an editor of the
standard to reflect data element model and
ISO/IEC 11179-5 naming conventions for
registry intended items
Standards continued
– SDMX – DDI NCubes were revised to incorporate the
ability to attach attributes to any area of a cube and
map cleanly into and out of SDMX cubes. SDMX has
added means of attaching fragments of DDI which
provide source and processing information that can
be indexed and delivered through SMDX tools.
– ISO 19115 (ANZLIC) – Geographic elements in DDI
are structured to reflect basic discovery elements
used by geographic search engines and provide the
detailed geographic structure information needed by
GIS system to incorporate the data accurately
DDI does not replace good
content
• DDI structures metadata to leverage content
–
–
–
–
–
Collection and processing
Discovery and access
Analysis and repurposing
Registries
Comparison
• DDI is not a software application
– Supports and informs software applications
• DDI is a neutral archival structure
– Preserving content and relationships
Value
• Supports consistent use concepts, questions,
variables, etc. throughout organization
• Supports implicit comparison through reuse of
content
• Supports explicit comparison by mapping
content between studies and to standard content
• Retention of explicit relationships between data
collection and the resulting data files
• Early capture of a broad range of metadata at
point of creation
Value - continued
•
•
•
•
•
Interoperability
Flexibility in data storage
Reuse of element structures
Strong data typing
Improved data mining between and across
systems
• Improved access to detailed metadata
DDI User Base
• Archives and data libraries worldwide
– Catalogs
– Data delivery
– Documentation delivery from data systems
• Research Institutes/Services Data Centers
– Documentation for data
– Data search and analysis systems
– Data management systems
• International Organizations and National
Statistical Agencies
– Data collection and management
Archives and Data Libraries
(examples)
• Catalogs
– ICPSR Data Catalog and Social Science Variable
Database
– CESSDA Data Portal
– The Dataverse Network (former Virtual Data
Collection)
• Data delivery
– California Digital Library “Counting California”
– National Geographic Historical Information System
• Documentation delivery
– Survey Documentation and Analysis (SDA)
– Data Liberation Initiative Metadata Collection
Research Institutes/Service Data
Centers (examples)
• Documentation for data
– German Microcensus (GESIS)
– Institute for the Study of Labor (IZA)
– US General Social Survey (NORC)
• Data search and analysis systems
– Nesstar
– Canadian Research Data Centres (RDC’s)
• Data management systems
– Questionnaire Development Documentation System
(University of Konstanz/GESIS)
Current DDI Products at ICPSR
• Most existing products currently in DDI 2.1 with
new additions moving to DDI 3
• DDI-XML variable-level codebooks output as PDF
files for downloading by users
• DDI-XML metadata records created initially by
data depositors and edited by ICPSR staff to
augment content and include additional fields
• Increasing use of DDI for special projects: Social
Science Variables Database, various
harmonization and data processing tasks
Potential Use of DDI 3 at ICPSR
• Information collected from data producers in precollection phase – Concept
• Metadata output from CAI applications – Data
Collection
• Processor‘s dashboard – Metadata Processing
• Metadata mining: New faceted search tool to
facilitate discovery through more precise
searching – Data Discovery
• Relational database for comparison and
harmonization across studies – Repurposing
Potential Use of DDI 3 at ICPSR
-2
• Use of DDI in combination with other metadata
standards, e.g., Dublin Core, MARC, PREMIS
• Beginning of FEDORA “object-centered“
implementation concepts into data processing
and data preservation strategies
• Processor‘s dashboard – Data Processing
• Relationships of study object to file object
• DDI 3 as “wrapper“ for all ICPSR metadata?
CURRENT WORKFLOW TECHNOLOGIES:
PROCESSOR-BASED
SPSS-BASED
NOT DDI-BASED
FUTURE WORKFLOW TECHNOLOGIES:
FEDORA DATASTREAMS
ICPSR “KEEPSAKE” OBJECTS
PREMIS METADATA
DDI “LIFECYCLE EVENTS” (PROCESSING HISTORY)
SSVD – The Public Search
• First batch of variable-level description files
uploaded into SSVD:
– Approx. 3,500 DDI files (one file per dataset),
representing
• Approx. 1,300 ICPSR studies (approx. 18.5 percent
of total ICPSR holdings, excluding US Census;
approx. 30 percent of holdings with data and
setups)
– Over 1,000,000 individual variable
descriptions; 23,000,000 categories
SSVD – The Public Search
• New database finalized Fall 2008
• Built to match DDI 3.0 data model
• Both DDI 2.x and DDI 3.0 compliant
– Designed to accept both DDI 2.x and 3.0
input and produce output in both versions
• ICPSR version currently uploads DDI
2.1 and generates DDI 3.0 individual
variable descriptions.
SSVD – The Public Search
Moving forward…
• Transition to automated DDI upload
– DDI uploaded at the time of study publication
– First quality check performed by study
processing staff
– Acceptable DDI immediately released for
public view
– Problematic DDI suppressed from public view
for further review, and upgrade as appropriate
Entry screen for internal search
Search results screen
IPUMS at MPC
• Did not use DDI because DDI 2 cannot
handle translation tables
• Currently in the process of mapping DDI 3
codebook output from IPUMS database
• Importing DDI 2 files from Microdata
Toolkit into processing, validation, and
harmonization system
NHGIS
• Contains historical aggregate data from
population, housing, agricultural, and economic
censuses as well as BEA data from 1790 to
2000
• Runs from DDI 2 nCube descriptions
• Searches variables, identifies related nCube
tables, determines geographic availability
• Generates data subsets with geographic links to
objects in NHGIS shape files, and shape files
Future Plans
• Funding has been obtained to improve search
and extraction system
• Current limitations of the system reflect
limitations of DDI 2
• Moving to DDI 3 will support:
– broader cross survey searching
– identification of common dimensions between
NCubes over time
– support harmonization instructions as well as
common transformations such as calculation of
medians
International Organizations and
National Statistical Agencies
• International Household Survey Network (IHSN)
–
–
–
–
Major international organizations involved
Coordination of activities
Adopted DDI 1/2.x as standard
Developed the Microdata Management Toolkit and related tools /
guidelines
– http://www.surveynetwork.org
• Accelerated Data Program (ADP)
– World Bank / Paris 21
– Implement IHSN activities in developing countries
• Task 1. Documentation and dissemination of existing survey
microdata.
– Has introduced DDI in national statistical agencies in over 50
countries
– http://www.surveynetwork.org/adp
INDEPTH/DSS Example
• 38 Demographic Surveillance Sites in 19
countries spanning Africa, South Asia, Central
American and Oceania
• Diverse yet similar health research portfolios
• Data management goals:
– Standardize and harmonize data collection
tools
– Cross-site comparability of information
– Sharing data effectively and efficiently
Reasons for choosing DDI
• “It will be ideal to describe our data for the
purposes of the Data Repository”
• “It has really powerful features that will
enable us to standardise several facets of our
work.”
• “I originally underestimated the usefulness
DDI will have as a means to harmonised data
collection between sites.”
• Ability to expand comparison and
harmonization with additional groups such as
AIDS research team
Statistical Agencies
• BLS considering publication of category
and coding standards supported by BLS
such as NAICS, SOC etc
• Statistics Canada considering publishing
concept schemes in DDI 3 for use by the
research community
• DDI is becoming more widely used for
survey and census collection in developing
countries (primarily Africa)
MQDS Version 1
• Extracted metadata from Blaise data
model as XML tagged data
• Provided user interface for selection of
– Blaise files
– Instrument questions and sections
– Types of metadata to extract
– Languages to display
– Style sheet for generation of instrument
documentation or codebook
Using MQDS V1 XML: Codebook in Five
Languages
National Latino and Asian American Study
www.icpsr.umich.edu/CPES
MQDS Version 1
• Limitations
– XML not DDI-compliant
• DDI Version 2 did not have XML tags for all
metadata provided by Blaise
• Did not provide easy means of adding XML tags
without becoming noncompliant
– XML files for complex surveys can be very large (text
files)
• Entire files had to be processed in computer
memory
• Limited ability to fully automate documentation
DDI Version 3
• Included extensions proposed by DDI
working group on instrument design
Persistent Content of Question
Question text
• Static
• Dynamic or variable
Multiple-part question
Response domain
•Open
•Set categories
•Special types (date, time, etc.)
Definitional text
Use of Question in Instrument
Order and routing
•Sequence / skip patterns
•Loops
Universe
Analysis unit
Instructions
MQDS Version 3
• Joint SRC and ICPSR venture
• Goals:
– Address version 2 limitations
• Process Blaise instrument of any size
– Exploit new elements and validate to the
recently released DDI version 3 standard
– Move from processing XML metadata in
memory to streaming metadata to a relational
database
MQDS Version 3
Relational Database: Import, Export,
Transform
SQL Server /
SQL Server Express
XML (DDI 3)
Relational
Db
Blaise
Datamode
l (BMI)
User specifies
input files
(location, file type,
etc.)
Blaise
Database
(BDB)
Other File
Types
(e.g. SAS,
SPSS,
etc)
Database
connectio
n settings
DDI 3
elements
not in
*.bmi
2.
Export
1.
Import
User specifies
output files
(location,
Language/locale,
XML output
options, etc.)
3.
Transform
Questionnaire
Codebook
User specifies stylesheet selection
criteria, type of output desired
(html, rtf, pdf), etc.
MQDS Version 3
• Relational database
– DDI compliant standardized tables
– Flexibility for SRC and ICPSR to add extensions that
meet their specific organizational needs
– Allows
• Automated documentation of any Blaise survey
instrument
• Importing and documenting data produced by
other software
• Lower cost development of other tools that
facilitate editing and disseminating data
<c:SubUniverse isVersionable="true" id="U4863" isInclusive="true">
<c:HumanReadable>-1</c:HumanReadable>
<c:MachineReadable>
<r:Code>GOSCHOL = Yes</r:Code>
</c:MachineReadable>
</c:SubUniverse>
<l:Variable id="V32373“>
<r:Name>A_FEM.AB_FEM.FMARIT</r:Name>
<r:UniverseReference>
<r:ID>U4657</r:ID>
</r:UniverseReference>
<l:QuestionReference>
<r:ID>Q69</r:ID>
</l:QuestionReference>
<c:SubUniverse isVersionable="true" id="U4657" isInclusive="false">
<c:HumanReadable>-1</c:HumanReadable>
<c:MachineReadable>
<r:Code>(MARSTAT = Widowed) OR (FMARSTAT =
Widowed)</r:Code>
</c:MachineReadable>
<d:ComputationItem id="CI150">
<d:Code programmingLanguage="Blaise">
<r:Code>FMARIT := 2</r:Code>
</d:Code>
<d:AssignedVariableReference isReference="true">
<r:ID>Q69</r:ID>
</d:AssignedVariableReference>
</d:ComputationItem>
Colectica Feature Overview
Current Focus: Data Collection
Survey Design: Diagram
• Visually design
survey instruments
• Drag items from
the toolbox
Colectica by Algenta Technologies
Survey Design: Item Editor
• Edit item details
using friendly input
forms
Colectica by Algenta Technologies
Multilingual Support
• All text fields can be represented in
multiple languages
Colectica by Algenta Technologies
Concept Repository
Concepts
• Use built-in or
custom concept
banks to describe
survey items
• Useful for
comparability
Colectica by Algenta Technologies
Question Repository
• Share questions
across studies
• Drag previouslyused questions or
sequences onto
new instruments
Colectica by Algenta Technologies
Import Existing CAI Code
• Import from:
– Blaise®
– CASES
– CSPro
• Support for additional languages can be
added
Colectica by Algenta Technologies
Generate CAI Source Code
• Currently support CASES
• Blaise® and CSPro coming soon
• Support for additional CAI systems can be
added
Colectica by Algenta Technologies
Generate Publishable
Documentation
• Generate codebooks and diagrams
• Output to HTML and PDF
Colectica by Algenta Technologies
Also: Study Concept & Design
• Basic support for Study Concept & Design
documentation
Colectica by Algenta Technologies
Generate DDI 3.1
Colectica by Algenta Technologies
Additional Information
• Beta available now
• Web: http://www.colectica.com/
• Email: [email protected]
Thank you
• DDI Alliance
– http://www.ddialliance.org
• Wendy Thomas
– [email protected]