Peter Henderson

I had my first experience last week of a useful conversation with an AI. I was trying to edit some text but was tired and struggling to find the right words. I gave ChatGPT an example of the style of writing I was aiming for, then asked it to summarise the text I was struggling…

I’ve been working with research for some time now but have always found it hard to fully understand p-values, confidence intervals and statistical significance. I think this is true of many (most?) people working in social science. I thought that writing a clear explanation in my own words could help. I’m posting my attempt here…

I’ve found it hard to find an intuitive explanation of what an odds ratio is and how it works. So here is my attempt: Odds ratios describe the relationship between two binary variables. They are often used in impact evaluation to show the relationship between exposure to an intervention (i.e. receiving the intervention vs being…

Stephanie Wardell at the EIF has a great piece summarising the state of the art on supporting evidence use. Anyone working at a What Works Centre or trying to have an impact through research should read it. My notes are below but it’s short enough that you should really read the whole thing. Stephanie has…

When can we learn from observational data? In 1965 Austin Bradford Hill created a set of criteria that are required for a confident claim that an observed link between an exposure and an outcome is causal. It’s designed for situations where we lack studies with a comparison group. I’m taking this from David Spiegelhalter’s The…