Incentive magnitude effects in experimental games: Bigger is not necessarily better, experimental data 2012-2017

Loomes, Graham (2019). Incentive magnitude effects in experimental games: Bigger is not necessarily better, experimental data 2012-2017. [Data Collection]. Colchester, Essex: UK Data Service. 10.5255/UKDA-SN-853538

Our project combines the expertise of nineteen established researchers in three major UK centres for behavioural research to conduct a timely and important programme of world class research. This research has two objectives: to develop and test cross-disciplinary models of human behaviour and behaviour change; and to draw out their implications for the formulation and evaluation of public policy.
At present, different social sciences provide very different accounts of the underpinnings and dynamics of human behaviour. At one end of a spectrum, conventional economic models are populated by highly rational individuals who make consistent and stable judgments among alternative choices. At the other end, many psychological models focus on cognitive mechanisms which may result in choices which are highly contextual and may often appear inconsistent. Both perspectives capture aspects of the truth, but considerable work is needed to bridge successfully between them. So our first objective is to create an integrated science of behaviour, leading to tractable and useful models, built from foundations which reflect well-established cognitive principles.
The second objective is to apply this new synthesis to policy. There has been growing interest in policy tools which respond to the realities of people's perceptions, judgements and choices. For example, the UK government's Behavioural Insight Unit draws on particular results from behavioural science to develop policies aimed at 'nudging' individuals towards 'better' decisions and policy goals. However, in the absence of a body of clearly articulated and properly tested theory which can replace the conventional underpinnings, these interventions risk being ad hoc, normatively ungrounded and may even turn out to be counter-productive. So we shall consider how the alternative models we develop can generate coherent and normatively defensible policy guidance.
In pursuit of our two broad and ambitious objectives, we will establish and operate an international, interdisciplinary, network to develop collaborative research, focussing on four inter-related themes. The first two explore questions of basic science; the second two draw out policy implications of that science.
1 - Understanding individual behaviour and behaviour change: The fundamental objective is to develop new and improved models of individual behaviour; to develop understanding of the extent to which values and preferences that underpin choice can be regarded as stable; alternatively how, or how much, behaviour can be changed or manipulated.

2 - Understanding social and interactive behaviour: We will investigate the real reasoning processes and values which shape peoples' choices in settings where they may (or may need to) take account of other peoples' actions or welfare. We will explore how material incentives, social and cultural factors, and individual differences shape motivations and strategic thinking.

3 - Rethinking the foundations of policy analysis: This theme reconsiders approaches to policy formulation, and the design of policy tools. Our aim is to rethink policy analysis building from richer and more realistic accounts of human agency informed by developments in Themes 1 and 2.

4 - Policy applications: We will trace through the implications of the basic science for policy via three applied projects each of which draws its primary motivation from one of Themes 1 - 3. In relation to Theme 1, we will use existing data sets (in collaboration with UK credit companies) to examine determinants of actual consumer credit behaviour. In relation to Theme 2, will use field experiments to investigate the formation of social values. In relation to Theme 3, we will explore strategies for evaluation of policies affecting health and safety.

Data description (abstract)

In experimental games, task-related incentives are payments to experimental subjects that vary according to their strategy choices and the consequent outcomes of the games. Limited evidence exists regarding incentive magnitude effects in experimental games. We examined one-off strategy choices and self-reported reasons for choices in eight 3 × 3 and four 4 × 4 normal-form games under task-related incentives of conventional magnitude and compared them with choices and reasons in the same games under incentives five times as large. Both strategy choices and self-reported reasons for choices were almost indistinguishable between the two conditions. These results are in line with earlier findings on individual decision making and with a parametric model, in which the incentive elasticity of effort is very small when compared with other factors, such as the complexity of the decision problem

Downloads

Altmetric