Thematic Analysis has become a justifiably popular, widely used qualitative analytic method across the health and social sciences, and beyond. The method is accessible, and deceptively simple. However this also means much can go wrong. In this talk, I connect to what we’ve identified as common problems in reported TA, to explore the crucial aspects for producing a good quality analysis and telling a good analytic story about the data and the analytic process.
In the leadup to the 2022 Australian election, there was considerable speculation about the possible success of minor parties and independents, and the likelihood of a hung parliament.
Just two weeks before election day, YouGov released estimates for all 151 House of Representatives electorates, the first time any organisation had done this during an Australian national election campaign. It indicated hung parliament speculation was wrong, predicting Labor would win 76-85 seats, with the most likely outcome being 80 (they won 77), and that the Coalition would win 58-68 (they won 58).
These estimates came from a model-based approach called Multilevel Regression and Poststratification (MRP), which combined a large survey with information from the Census and other data, allowing for a more granular approach to polling.
YouGov’s Director of Data Science — Public Affairs and Polling, Dr Shaun Ratcliff, will talk about how the model worked, what it said about the election, and lessons from YouGov’s innovations at this election.