Via Paul Krugman, a paper, finding that most pundits do no better at making predictions than a flipped coin. Remarkably, the piece fails to cite the ground-shattering analysis which arguably launched the field of quantitative pundit studies, Lee Sigelman et al.’s statistical analysis of the predictive powers of the McLaughlin group. Sigelman and colleagues find that McLaughlin predictions were only right 50.1% of the time – it would seem that there has not been much improvement in the meantime (to the extent that the recent paper’s findings are reliable – I wouldn’t be staking me life on them myself).
{ 3 comments… read them below or add one }
Do bear in mind that pundits are asked question specifically because those question are difficult. Plumbers answering question would be more accurate, I hope, because of the kinds of questions they would be asked.
To put it another way, McLaughlin forms his questions so that his group will be evenly divided upon them.
i do the same thing with my plumbers.