Unlocking the power of Administrative Data Yi-Ping Tseng Economic - - PowerPoint PPT Presentation

unlocking the power of
SMART_READER_LITE
LIVE PREVIEW

Unlocking the power of Administrative Data Yi-Ping Tseng Economic - - PowerPoint PPT Presentation

MELBOURNE INSTITUTE Applied Economic & Social Research Unlocking the power of Administrative Data Yi-Ping Tseng Economic and Social Outlook Conference Tuesday, 25 July 2017 Page 1 Birth Death Child records records Protection Police


slide-1
SLIDE 1

Page 1

MELBOURNE INSTITUTE Applied Economic & Social Research

Unlocking the power of Administrative Data

Yi-Ping Tseng

Economic and Social Outlook Conference Tuesday, 25 July 2017

slide-2
SLIDE 2

Page 2 Unlocking the power of Administrative Data | Tuesday, 25 July 2017

What is ‘administrative data’?

  • Collected for operational purposes
  • Provides records of:

services (or events, transactions, etc.) and people receiving (& providing) them

Administrative data Birth records

Child Protection

Education Schools Admin NAPLAN

Centrelink Payments

Homeless service Employment service

Tax records

Hospital

Medicare

Police records

Death records

slide-3
SLIDE 3

Page 3 Page 3 Unlocking the of Administrative Data | Tuesday, 25 July 2017

Usi Using ng administr administrati tive e da data f ta for

  • r rese

esear arch h

Advantages Disadvantages

Cheaper Requires more cleaning and clear documentation

  • f business rules.
slide-4
SLIDE 4

Page 4 Page 4 Unlocking the of Administrative Data | Tuesday, 25 July 2017

Usi Using ng administr administrati tive e da data f ta for

  • r rese

esear arch h

Advantages Disadvantages

Cheaper Requires more cleaning and clear documentation

  • f business rules.

More accurate in terms of: Timing and detail of services provided Verifiable information (eg. identity) Intentional misreporting may be more prevalent

slide-5
SLIDE 5

Page 5 Page 5 Unlocking the of Administrative Data | Tuesday, 25 July 2017

Usi Using ng administr administrati tive e da data f ta for

  • r rese

esear arch h

Advantages Disadvantages

Cheaper Requires more cleaning and clear documentation

  • f business rules.

More accurate in terms of: Timing and detail of services provided Verifiable information (eg. identity) Intentional misreporting may be more prevalent Larger sample size (population of service users) More statistical power More scope for sub-group analysis No information for non service users Limited information on background characteristics Limited information on destination of ‘service exits’

slide-6
SLIDE 6

Page 6 Unlocking the power of Administrative Data | Tuesday, 25 July 2017

Administrative data Birth records

Child Protection

Education Schools Admin NAPLAN

Centrelink Payments

Homeless service Employment service

Tax records

Hospital

Medicare

Police records

Death records

Da Data ta li linka nkage ge:

The more that administrative data sets are linked, the greater the scope to understand:

  • the behaviours of service users
  • the interaction among service

systems

slide-7
SLIDE 7

Page 7 Page 7 Unlocking the of Administrative Data | Tuesday, 25 July 2017

Usi Using ng administr administrati tive e da data f ta for

  • r rese

esear arch h

Advantages Disadvantages

Cheaper Requires more cleaning and clear documentation

  • f business rules.

More accurate in terms of: Timing and detail of services provided Verifiable information (eg. identity) Intentional misreporting may be more prevalent Larger sample size (population of service users) More statistical power More scope for sub-group analysis No information for non service users Limited information on background characteristics Limited information on destination of ‘service exits’ Continuous data collection and ease of extraction (up to date information)

slide-8
SLIDE 8

Page 8 Page 8

Administr Administrati tive e da data f ta for

  • r po

poli licy an y anal alysis ysis

  • A. Policy/Program Evaluation
  • Using advanced ‘quasi-experimental’ econometric techniques
  • Used in conjunction with experiments (Randomised Control Trials), administrative data are

particularly valuable for

  • Analysing long term impacts
  • Tracking hard-to-reach participants
  • Cost Benefit Analysis
  • Process evaluation (data from delivery of intervention, eg. case notes)

Unlocking the power of Administrative Data | Tuesday, 25 July 2017

slide-9
SLIDE 9

Page 9 Page 9

Admin Administr istrativ tive e da data f ta for

  • r po

poli licy an y anal alys ysis is

  • B. Identifying a need for reform
  • Monitoring trends, outcomes
  • Utilise longitudinal aspect of the data
  • E.g. What proportion are heavy users (or frequent users)? Are these proportions changing over

time? Or across cohorts?

  • Analysing and predicting service use patterns
  • Using advanced techniques such as “machine learning”

Unlocking the power of Administrative Data | Tuesday, 25 July 2017

slide-10
SLIDE 10

Page 10 Page 10

Administr Administrati tive e da data f ta for

  • r po

poli licy an y anal alysis ysis

  • C. Use in conjunction with survey data to study behaviour
  • Linking administrative data to general survey data
  • Can provide accurate service usage information
  • Eg. take-up rates and patterns
  • A sampling base for specialised surveys
  • Use administrative data to find target population and conduct surveys
  • Eg. Journeys Home Survey uses Centrelink data ‘homeless’ flag to define target population

Unlocking the power of Administrative Data | Tuesday, 25 July 2017

slide-11
SLIDE 11

Page 11 Page 11

Unloc Unlocking king t the he po power er of

  • f administr

administrati tive e da data f ta for

  • r

evide videnc nce-ba base sed d po poli licy

  • Address some limitations of administrative data
  • Promote data linkage
  • Improve quality of data collection (e.g. promote its importance to front line staff; set protocols)
  • Transform existing administrative data to suit research purposes
  • Take research needs into account when designing future collections
  • Make data more accessible for policy research

Unlocking the power of Administrative Data | Tuesday, 25 July 2017