Snap judgement polling: street interviews enabled by new technology

Katherine Anderson, Malcolm Wright and Meagan Wheeler

Introduction

Election polling dates back to 1824, when the Harrisburg Pennsylvanian newspaper asked some residents of Wilmington, Delaware, which presidential favourites they preferred, using a simple straw poll (Smith 1990). Since then election polling has evolved. One strand of this evolution has been improved sampling and questioning approaches. Communication technologies have also resulted in new collection methods. The successive introduction of mail, telephone and now online surveys, has enabled polling organisations to develop new polling methods and collect data more efficiently.

Recently, internet-enabled mobile phones have made it possible to conduct very sophisticated surveys with respondents randomly recruited in public places. We capitalise on these technological developments to test a new method for pre-testing political candidates: showing naive respondents photographs of candidates’ faces on iPhones and asking them to make snap judgements about the competency of the candidates. We test the effect of including rational information like candidates’ political affiliations on the predictive accuracy of judgements as well as comparing the accuracy of different question types.