Can search engine advertising help access rare samples?

Daniel Nunan and Simon Knox


Technological changes through the ‘noughties’ have enabled a huge increase in the use of web-based surveys by researchers. Trends over the last decade have been driven by the adoption of the web and mobile devices as dominant forms of communication and information distribution. In many countries, web access can now be considered as ubiquitous as fixed-line telephony once was (Geoghegan 2009). At the same time, technology has also challenged existing modes of data collection. For example, increasing use of caller ID devices has substantially increased non-response rates for telephone research (Jarvis 2002). Despite the popularity of online surveys there have also been increased concerns over survey error, particularly around selection and non-response bias (Sharot 2010). Yet, online surveys have also been seen as an effective means of reaching empirically underrepresented and hard-to-reach populations (Koch & Emrey 2001). The last few decades have seen a reduction in demand for general samples of consumers, and an increase in research using highly specific groups, or ‘rare populations’, that represent a small proportion of an overall population (Sudman & Blair 1999). Reaching such populations through traditional random sampling techniques may require significantly more resources than are available to build a sufficient sample size. In these circumstances, where samples would otherwise be unavailable, pragmatism forces the use of more purposive, and perhaps less desirable, sampling techniques.