Presentation is loading. Please wait.

Presentation is loading. Please wait.

Implementing DDI in a Survey Organisation

Similar presentations


Presentation on theme: "Implementing DDI in a Survey Organisation"— Presentation transcript:

1 Implementing DDI in a Survey Organisation
10/05/2019 Implementing DDI in a Survey Organisation 3 December 2012

2 10/05/2019 Contents Background Our aims The transition

3 10/05/2019 Background 1.

4 10/05/2019 Background NatCen Social Research is an independent social research organisation in the UK Carry out large scale surveys from design to reporting Health Survey for England and Scottish Health Survey British Social Attitudes Survey Growing Up in Scotland English Longitudinal Study of Ageing Understanding Society Make data available for secondary analysis through the UK Data Service Produce user documentation that allows data users to understand the data

5 Documentation User Guide
10/05/2019 Documentation User Guide Background to study Sampling Survey methodology Weighting Response rates Documentation of survey instrument showing question text, answer categories, routing and variable name Information on transformations carried out on the data, derived variables, recode Dataset documentation – variable lists arranged by topic (All documentation is provided as PDF files)

6 Documentation – other issues
10/05/2019 Documentation – other issues Some studies involve collecting biomarkers (blood, saliva, urine), separate self-completion data Longitudinal studies also need to document changes between waves Other types of data collection, audio files, social media etc. Currently this information is compiled as text from various sources But all this is metadata and needs to be managed and made accessible

7 Current working patterns
10/05/2019 Current working patterns Studies run by research groups within the organisation Larger projects have data managers, in smaller projects, data management is done by researchers Each group evolved their own way of producing documentation – often last thing done in the project Metadata ‘hidden’ in text/pdf files, sometimes not easily found for re-use After recent reorganisation we are reviewing how we do data management, including capturing and managing metadata

8 10/05/2019 Our aims 2.

9 10/05/2019 Aims Move away from the situation where metadata ‘capture’ is done at the end of a study by someone who may not have been involved in the stages they are documenting Metadata ‘capture’ will be done at the stage the information is first generated AND this should not be an extra burden for anyone Final documentation will be produced using the existing metadata from each of the earlier stages

10 10/05/2019 How will we do this Not necessarily looking for an off the shelf complete study package We want to: exploit capacity of our existing software and processes to capture/export metadata find opportunities to introduce new tools that enhance our current process

11 10/05/2019 How will we do this We believe that metadata tools are useful/acceptable if: they provide a straightforward, structured way of capturing and communicating information they offer some benefit at the capture stage the information provided can be easily extracted later

12 10/05/2019 Transition 3. How do we get from where we are now to where we’d like to be

13 Existing flows (illustrative)
10/05/2019 Existing flows (illustrative) CAPI Blaise Study design Question design Survey implementation CAWI Word Word Self completion Word / pdf Word, Blaise Coding / editing Data processing SPSS FW monitoring Newfield Data delivery SPSS Most projects this whole cycle. 1 study design (various, mainly word, some / xl) researchers->collaborators/clients 2. Q design (various – mostly word, some other – some copying from previous, some direct into blaise) 3. Implementation in some mode(s) (blaise / word->pdf) 4. Fw monitoring / reporting (newfield SQLs / xls / other) 5. Coding (blaise / qdcor)(word spec) also 1st stage cleaning / some DVs / reconciliation 6. Data processing (SPSS) 2nd stage cleaning / DVs / weights / labels 7 data delivery (SPSS / SAS / Stata) 8 archive (SPSS) Documentation from all these places Could add to this – project management, other things. Automatically Up till recently have been thinking about CAI instrument and dataset as metadata source Now looking at beginning of process and throughout the cycle Documentation Word / pdfD Archive SPSS Other sources Publications, citations, etc.

14 Stages where a new tool would be welcomed
10/05/2019 Stages where a new tool would be welcomed Questionnaire design Coding and editing Project database Stages where a new tool would be welcomed Where it would provide a structure way of communicating information. Where there would be a tangible benefit to team AND where the information provided can be extracted as metadata. Currently identified 3 possible ‘entry points’ DON’T TALK ABOUT THEM YET – NEXT 3 SLIDES Q design Coding/editing spec Proj database / proj design initiation

15 Questionnaire design stage
10/05/2019 Questionnaire design stage Researcher works with client to define and test questions Researcher passes final question specification to programmer Programmers would like a consistent structured way for this to happen. Questionnaire design stage – communication between researcher and programmer Researcher works with client to define and test questions or choose validated questions to reuse from other studies. Browser interface on our server for client to access??? Researcher lets programmer know which questions to add to CAPI programme or to reuse from other studies. Usually done by researcher sending Word draft and requesting changes by or Excel ‘log’. Varies across projects. Programmers would like a consistent structured way for this to happen. Researcher types the text anyway. So ideal place to introduce software that allows question specification and can export DDI3 XML

16 10/05/2019 Coding and edit Researcher produces list of questions that will need coding with details and instructions Could be replaced with a structured document Coding and edit sta Researcher produces list of questions that will need coding and provides code frames (or for standard questions like occupational coding, coding team have existing code frames. Again currently done as Word document. Could be replaced with a structured ‘document’ where field are completed by researcher and DDI3 XML exported ge – communication between researcher, data manager, programmer and coding team

17 Project database – administration
10/05/2019 Project database – administration Kept by our administrator, with basic information about each project Fields could be added to provide a wider range of metadata Project database – administration Kept by our administrator. Basic information about each project. Project number, name, research director in charge, fieldwork manager, client. SQL database ? Fields could be added to provide a wider range of metadata – elements of project design Could provide the basis for a metadata repository or feed into it.

18 10/05/2019 Thank you If you want further information or would like to contact the authors, Joan Corbett Senior Data Manager E. Roger Stafford E. Visit us online, natcen.ac.uk


Download ppt "Implementing DDI in a Survey Organisation"

Similar presentations


Ads by Google