If you're running randomised trials, the linear probability model is your friend. At BETA, we use the linear probability model with robust standard errors as our primary estimator.

BETA recently contributed to the OECD’s LOGIC report, which identifies effective practices for incorporating behavioural science into mainstream policymaking.

Within the Australian Public Service, we're lucky to have a highly active and engaged community of behavioural insights practitioners – The Behavioural Insights Practitioners Network (BIPN).

This is Part 2 of our blog series on how we used machine learning to analyse gendered language in job ads. In this blog, we go through the technical nitty gritty of how BETA analysed 12 million job ads, and what we found.

Thanks to the release of OpenAI’s ChatGPT, we’ve all become aware of the power of natural language processing (NLP). But what is NLP? BETA explores the common terms and make sense of some of the hype.

2023 was a time of change and excitement at BETA. We started the year with big news, and wrapped up with big data; several of our projects related to workforce issues, including disability and return to work; and we continued to engage with the big issues in behavioural science, connecting with colleagues across the APS and beyond.

How can academics engage with government? BETA is supported by our Academic Advisory Panel (AAP) who provide expert advice on the application of behavioural economics in public policy. In this blog, we share four key tips for academics engaging with government.

How does BETA ensure it applies behavioural insights ethically? Our blog explores the key principles to guide the ethical application of behavioural science in government.

Need to consult the public? BETA shares some lessons from a research approach that can be applied to all kinds of community engagements.

Regulators are increasingly recognising the value of behavioural economics in regulatory design and enforcement. And in 2023, it’s clear that the value of the behavioural perspective extends well beyond nudges.