• subscribe

    • landline

    Mind the ‘credibility’ gap

    Colby Cosh finds out what subsets, modelling assumptions and ‘non-probability samples’ have to do with polling these days