Internet surveys are attractive because they are efficient and cost effective, but they do have their drawbacks. Internet surveys in the UK are based on “opt-in” non-probability samples. Unlike traditional probability samples that rely on the principle of randomness, such non-probability techniques rely on targeted advertising campaigns, monetary incentives to recruit participants, and quotas. Whatever weights and matching strategies may be applied to such panels, they may not be as representative as a probability sample. Probability-sample based internet surveys have been less common, although with very high internet penetration in many countries they deserve much more investigation. Like postal surveys, internet surveys rely on self-completion, but which do respondents choose to fill out when given both options? We examine these questions comparing responses on internet surveys to more traditional face to face or mail back surveys using data from the British Election Study and the New Zealand Election Study.
