1. Ibase in Manchester
Corporate Research & Intelligence Chief Executives

2. Manchester’s iBase Journey
Commitment to data and evidence – spreadsheet solutions
Building Big Data solutions to support PSR – database solutions
Active use of data – Strategically & Operationally
2010
2011
2012
2013
2014
2015
2016
2017
Community Budgets Programme
National Troubled Families Launch
Ibase
Proof of Concept
Ibase
Phase 1
Date schema design, build, testing, and Information Governance
Ibase
Phase 2
Rollout to Front Line as part of Troubled Families Phase 2
ibase Phase 3 – Automation, Wider Roll Out & Reporting
Impact of iBase
Performance & Identification
Analysis & Testing
Evaluation & Insight

3. What we built FAMILY ADDRESS LEVEL 1 LEVEL 2/3 PERSON LEVEL 2 LEVEL 3
CTAX
BASED ON:
1 – RCTX
2 – MICARE RELATION
3 – HOUSING BEN / FSM
4 – SURNAME & UPRN
LANDLORD
FAMILY
ADDRESS
ASB CASELOAD
INTERVENTION
LEVEL 1
LEVEL 2/3
POLICE CALLOUT
PERSON
IBASE PERSON ID
NAME & DoB
MICARE ID
PNC ID
PUPIL ID
NHS No.
CLAIMANT ID
NINo
PLACE FACTORS
BASED ON:
LLPG UPRNs
MICARE PREVIOUS ADRESSES
DE-DUPLICATED IDs AND UPRNs, WITH STANDARDISED / COMMON NAMES, DoBs, & ADRESSES
LEVEL 2
HEALTH
SCHOOL
OFFENDING
FINANCIAL
WORK & SKILLS
SAFEGUARDING
DRUGS
ALCOHOL
MENTAL HELATH
SCHOOL ROLL
ABSENCE
CME
EXCLUSION
PRU
SEN
OFFENCE
ASB LEGAL
ANY ADDRESS ISSUES WITH ASB OR CALLOUT
CTAX ARREARS
FSM
HOUSING BEN
NEET
DWP
DWP BENEFITS
MiCare Master
CIN
CPP
LAC
OTHER SW?
LEVEL 3

4. Using iBase Operationally
Case Conferences
Targeted Approach
Telephone Calls
Case Reviews
Intelligence Led
Interviews / Assessments

5. 1. Find the Case / Family

6. 2. Direct Links to Case

7. 3. Immediate Family

8. 4. Extended Family

9. 5. Exploring the Issues

10. 6. Looking for Links

11. 7. Expanding the Links

12. 8. Finding the Connection

13. Using iBase Strategically
Family Focused
Person Focused
PSR
Work & Skills
Criminal Justice
Health & Social Care
Complex Families
Early Years
Growth & Neighbourhoods
Provider Focused
Place Focused
Our evaluation approach needed to respond to a rapid shift in the type & scale of services being offered to users, and needed to respect all of the interdependencies
Crucial to avoid a trial & error approach but rather focus on evidence based solutions and use a combination of evaluation methods to inform both strategic and operational decisions, some of which are evolving.
Using operational insight alongside data patterns provide a timely addition to evaluation

14. What are the Research Questions?
Understanding the impact on the system (so it can be planned)
Understanding the causal factors behind the outcomes (so we provide the right service)
Predictive modelling
Ultimately leading to decision making tools

15. Performance & Tracking Who and What (single issues)?
Different Issues
Different People
More Concern for Safety, Domestic Violence, Mental Health, & Safeguarding
More Young Families and fewer Single Adults
Different Responses
2,459 Referrals
1,951 (79%) Allocated to Tier 1
1,661 (68%) Begin Intervention
145 (6%) Disengage
1,516 (62%) Complete Intervention
290 (11%) Deemed Inappropriate
508 (21%) Signposted to other Services
32% of referrals now going to alternative interventions

16. Profiling - Sequence Analysis Who and What (multiple issues)?
Figure 1 – this represents a subset of all family records, a purple colour indicates the presence of an event (e.g. receiving housing benefit, or having school absence events) whereas a turquoise colour indicates the absence of such events. c.80,000 records
Figure 2 represents just the TF data (with the extra attribute of whether they have received any intervention treatment), this shows far more events occur for these families. c.3,000 records
Engaged families with no CIN
Engaged families with complex mix of issues
Families with Housing Benefit, Free School Meals, CIN, Absence, Exclusions, and Social Renting
Families with Absence, Exclusions, Social Renting and some CIN
Not Engaged but with complex mix of issues

17. Places & People – Spatial Analysis Where and Links?
Location of Troubled Families
These areas are similar with regards of Median Income, Types of problems the TFUs have, Deprivation indices
But they differ in: Transience, Predominant tenure
Density of Troubled Families
TF scattered across all the city. Very few areas with no TFUs (Didsbury). Number of TFU never exceeds 5% of all households in any LSOA
Substantial clusters: Harpurhey, Gorton / Belle Vue, Whythenshawe

18. Causality & Links – Decision Trees
Why?
CIN Events
More that 3 CIN Events
Less than 2 CIN Events
No ASB Issues
Not Claiming Housing Benefit
Good Attendance
No Free School Meals
Social renting <50%
Social Renting >=50%
Free School Meals
Persistent Absence
Claiming Housing Benefit
Has ASB Issues
More that 2 CIN Events
No Events
Social Renting >=18%
More that 1 ASB Issue
Single ASB Incident
Social Renting <18%
We probably know this
But we don’t know these links
Reviewing recent history and using the common decisions to predict a likely future decision

19. Predictive Modelling Changing the systems
Using data mining to identify your clusters / cohorts
Testing cluster / cohorts characteristics to identify the significant factors
Apply the significant factors to the whole population to identify scale
Use decision trees and sequence analysis type tools to test likely impact of decision for this broader cohort
Use this to inform the business planning
Identify a cohort / issue
Review the Existing Evidence
Investigate the Characteristics
Identify the Key factors
Look for Similar Groups
Inform Business Planning

20. What next? Building performance reporting tools
Building threshold and alerting tools
Roll out to more users (Social Workers, CSE teams, Locality Teams, etc)
Connecting more data, stakeholders and places

21. Any questions or comments please contact: Paul Holme Research and Intelligence Manager Lee McNichol Senior Research Officer