Page 394 - Special Topic Session (STS) - Volume 2

STS507 Vince G. et al.
 In the first research of its kind, researchers have been able to describe the
whole story about what it means for NZ babies born prematurely. and their
long term outcomes.
 Oranga Tamariki are gaining insights from the IDI that will help improve
the well-being of children and young people.

Trying to use large scale models to help develop Policy is still a work in
progress
The most striking lessons have arisen from participating in the attempts
by major policy agencies to realise value from making substantial investments
in microsimulation. The approach of the Justice Sector, for example, has been
shaped by the view that social outcomes are largely emergent, in the sense
that the causal mechanisms cannot be fully specified. This thinking is
influenced by theories which found that integrated theories tend to explain
less variance than individual theories. Consequently, their microsimulation
model is quite simple and designed not to explain causal mechanisms but to
describe baseline trajectories of offending. The IDI is primarily useful in terms
of understanding the distribution of risk across the population, and patterns
of service use and co-occurrence of problems. This helps understand who the
Justice Sector might want to focus on and which government services they're
already interacting with.
In broadly similar vein, the Vulnerable Children’s model aims to understand
demand for support services, the benefits and costs of providing services, and
the key drivers of people transitioning into, through and out of the support
systems. The idea is broadly that in the current timeframe data can be
collected on subjective measures of well-being and detailed segmentation
analysis of the operation of the current system can be done, while a longer
term model is used to understand whether there is a relationship between
particular indicators of wellbeing and future outcomes.
Both these approaches are responses to the general problem of trying to
predict how people will respond to different incentives and constraints. This is
more feasible in the short term and for small changes in programmes but
inherently difficult for significant life changes over an extended period of time.
The problem is trying to find practical ways of getting a sense of where
these limits of what can usefully be predicted turn out to be in practice. Short
term projections of service usage can be predicted from data about upstream
pressure within the system, but it can be surprisingly hard to make accurate
predictions about major service usage over a 4 year time horizon.

383 | I S I W S C 2 0 1 9