ESRA 2027 sessions by theme

Back to Overview of Sessions

Session Details

Political polling has repeatedly adapted to changes in how surveys are conducted and how electorates behave. The growth of online and non-probability recruitment, declining and increasingly selective participation, and more complex patterns of political engagement have been met with advances in weighting, modelling and survey design. At the same time, more fragmented party systems, electoral volatility, polarisation and changing patterns of turnout can make the relationship between survey respondents, the wider electorate and eventual voters harder to characterise. These developments do not invalidate established polling methods, but they create new challenges for how samples are constructed, adjusted, validated and translated into estimates of electoral behaviour.

This session invites contributions exploring what the next polling paradigm might look like. Rather than seeking a single replacement for established methods, we welcome research that strengthens, extends or complements the current polling toolkit. Topics may include:

• Advances in MRP, including diagnostics, model specification and implementation, and improved or synthetic poststratification frames.
• Approaches to non-ignorable nonresponse, including selection models, sensitivity analysis, design-based interventions, and methods for understanding residual within-cell differences
• Improvements in weighting and adjustment, including methods to better select weighting variables, the use of non-census variables and synthetic population frames, weighting approaches beyond raking, and probability/non-probability sample integration.
• Better modelling of electoral turnout, including probabilistic likely-voter models, voter-file information, longitudinal validation and approaches to adapt to changing electorates and party systems.
• Greater sensitivity analysis and transparency around nonresponse, measurement error, benchmark uncertainty, model dependence and other error sources from the TSE.
• Approaches enhancing or substituting surveys, including election forecasting, prediction markets, or synthetic data.

The aim is to bring together competing and complementary approaches to how polling can continue to evolve as both the political environment and the methodological possibilities for measuring and predicting.