Page 1 of 2
Key Issues in Internet Research Written by John Aitchison from First Line Research
Internet based research continues to grow in popularity due to high and increasing levels of healthcare professional (HCP) internet penetration and usage; an ever broadening range of applications; plus a steadily improving track record. Across all the EU Member States an average of 65 per cent of primary care physicians (PCPs) now use the internet either from home or from work. For the ‘big five’ EU markets the stats improve: in the UK and Germany, the proportion is over 90 per cent; in France, approximately 75 per cent; and in Italy and Spain approximately 70 per cent. Typically, levels of internet penetration amongst hospital specialists are 5 to 10 percentage points lower than for PCPs. Nevertheless, the methodology still has its sceptics who argue that behind every opportunity lies a pitfall. This brief analysis offers the reader a practical overview of internet research within our sector.
There are dozens of online HCP ‘panels’ serving the UK market. In reality some are much larger and/or more ‘representative’ than others. A handful are close to being truly managed panels, whilst others are better described as opted-in databases, or just simple lists. In many cases there may be a question mark over the extent to which HCPs most likely to take online surveys (i.e. higher-end internet users) are attitudinally different to the norm. HCP samples for adhoc online surveys are therefore often best sourced via a combination of established healthcare panels, and telephone recruitment.
Most quantitative research projects, whether adhoc or syndicated, can be administered online. These days, technical and sampling obstacles rarely stand in the way and the biggest question is usually, ‘Can the objectives be achieved using a self-completion survey?’. Consequently the list of applications is growing, along with HCP enthusiasm for the approach. Product and concept tests; advertising and logo tests; customer surveys; usage, perception and behaviour surveys; diary and tracking studies; website evaluation; omnibus; conjoint; market segmentation; and so on – all have been successfully conducted online. GP Omnibus services that can turn questions into answers within 48 hours are now par for the course.
Whilst there are nuances to consider, design freedom is rarely hampered by things that ‘can’t be done’. For example, there is a persistent myth that detailed images shouldn’t be incorporated into surveys because they are likely to cause unacceptably lengthy page download times. In reality an A4-size Ad execution can be clearly and cleanly reproduced on screen in less than five seconds over a 56k modem connection, so long as the correct file format is employed.
More challenging are the grittier tasks - scripting; entering and checking validation procedures; and creating routing, piping, quotas, screen-outs and other survey logic. Whilst the fruit of this labour is a survey that is easy on the eye, intuitive and yields exceptionally accurate output data, getting there takes as much skill and patience as would the preparation on a face to face or telephone study. Online studies are often ‘quicker’ than offline equivalents, but squeezing lead-in times can be perilous.
Whilst almost all quantitative question variations can be scripted, people often wonder about how well open questions work online. Whilst there is no definitive answer, the general observation is that questions of the, ‘other, please specify’ variety work well in sensible numbers and genuinely engaging interrogative open questions can work very well when used sparingly. However respondent fatigue sets in rapidly when blander, “Why do you say that?” style questions crop up after every rating scale.
High-end survey software now supports all major, and most minor, languages and – in countries where internet penetration and usage support an online approach – online surveys can offer administration, cost and survey quality benefits. Translation quality and proof-reading is especially important, since the self-completion methodology leaves no room for respondent doubt.