Using Data to Improve Public Services and Policy

1

Using Data to Improve Public Services and

Policy

Tom Symons - Deputy Director, A Fairer Start

Rachel Wilcock - Data Scientist, A Fairer Start

30th September 2021

Today’s Session

● About Nesta

● Why use data?

● Fundamentals of a good data project

● Applying data to the early years

● Lessons for good data use in government

● Data scientists - attracting and retaining

2

About Nesta

3

Nesta’s Strategy 2021-26 - Three Innovation Missions

A Fairer Start A Healthy Life A Sustainable

Future

Why use data analytics in

government?

Data can help us do many things

● Communicate

● Learn

● Be transparent

Today’s focus: data to help us

make better decisions

Old approach; new technologies

In 1854, John Snow plotted the

location of deaths in London’s Soho to

show that a cholera outbreak was

caused by contamination of a local

water pump. The map he created led

to the pump handle being removed,

saving many lives.

Today computers enable us to

analyse greater quantities of data in

more sophisticated ways.

12 London boroughs and the Greater

London Authority analysed data on the

known risk factors associated with

unlicensed Houses in Multiple

Occupation (HMOs) to help building

inspectors find other properties that were

likely to be unlicensed HMOs. Its aim was

to increase licence revenues and

protect vulnerable tenants.

Targeting HMO building inspections

8Image: AD_Imgaes | Pixabay CC0 Creative Commons

In New Orleans, data on the nature,

location and timing of past emergencies

was analysed in order to predict where

and when future emergencies could

happen.

This analysis helped identify the optimum

places to park ambulances on standby

to reduce response times.

Optimising ambulance standby locations

Image: Emergency Medical Transport, Inc.

9

The government’s data team worked

with telecoms companies to analyse call

data to see where mobile calls were

made during the hours of midnight and

5am. The companies matched this to the

billing address of the callers, anonymised

and aggregated the data, and then

passed it on to the city government. This

data was analysed to find the routes of

expected heaviest demand for a night

bus.

With just 30 buses, they’ve met nearly

half the demand at night.

Seoul Night Bus

10

Fundamentals of a good

product

Can I address this problem / answer this question with data?

Specific Problem

Defined Action

Data Product

Available Data

Start here

But don’t forget to evaluate

it…

And be prepared that not

everything will work

We ran an RCT which found the

prioritised list did not result in a higher hit

rate...

This is a useful finding!

Subsequent work in LB Barking and

Dagenham has improved the accuracy

of the algorithm

Targeting HMO building inspections

Image: AD_Imgaes | Pixabay CC0 Creative Commons14

Nesta’s Fairer Start Local

Programme

Using data to improve early

child development outcomes

A fairer start

We want every child to have an equal start in life.

Our mission is to narrow the outcome gap between children growing up in disadvantage and their peers.

April 2018-March 2020 data

Take-up of Health Review by geographic area

The problem: not all parents engage with the HCS. Where a child is born in York is key indicator of how likely they are to receive the Health Review

Case Study - building a data dashboard for York to improve take-up of the Healthy Child Service

The action: Target proactive outreach in the geographic areas where take-up is lowest

Case Study - building a data dashboard for York to improve take-up of the Healthy Child Service

Case Study - building a data dashboard for York to improve take-up of the Healthy Child Service

● Dataset was simple, clean and accessible

● Their IG team were responsive, experienced and worked closely with us.

● Close links with the domain experts to give us the insights as to what is most useful.

● Link with open datasets to enrich

Benefits:

● Real-time data to help monitor need, screen for risk

& target resource

● Accessible & easy to use

● Combines quantitative data with qualitative insights

Next steps:

● Build capacity to enable York staff use the

dashboard in their day-to-day work

● Integrate with other databases

● Evaluate impact on service take up

Building a

data dashboard

for York Healthy

Child Service

Principles of good data use

Key learnings about running good data projects

1. Get senior buy-in

1. Be flexible (or agile)

1. Data sharing takes time!

1. Be receptive

1. Consider ethics

Data Scientists - attracting

and retaining....

Insights from the Data Analytics Practice at Nesta

Data Science includes a wide range of

skill sets, and having a long list of these as

“essential” on job adverts is off putting

You get to work with interesting datasets and feel like you are helping at a local level

Data science in some local governments is a new addition which is both positive and negative

Forge links with other

LAs and share learning

experiences