Surveys are vital tools used throughout business, government, and science to inform decision and policy makers. The critical step in any good survey project is planning. The most sophisticated surveying tools, most experienced interviewers, and largest sample sizes can all become irrelevant when a survey has been poorly designed. Allocating insufficient budget to the planning stage can lead to inefficient or wasteful use of resources, costly mistakes, and poor results. One important role of the survey researcher is to balance budget constraints with the objectives of the survey and find a cost efficient means to collect data with acceptable assurance.
The advent of social media as a tool for both recruiting respondents and delivering the survey introduces new challenges to survey planning. A number of survey software companies offer survey tools for social media platforms, marketing them as a quick and cost effective way to reach target audiences. While this is potentially great news for survey researchers, it is important that the inherent dangers of this medium are understood.
Senator Malcolm Roberts learnt the dangers of poorly designed Twitter polling earlier this year when he asked, “How should the government spend $5 billion of taxpayer money currently used on green energy?” He was convinced the poll had been rigged when 87% of replies stated “green energy” rather than the three alternative options he presented of “paying off debt”, “infrastructure” or “education/health”. Was the poll an accurate reflection of popular opinion? Who knows? What is clear is he did not understand the limitations of the Twitter platform, nor the audience his Tweet would have reached.
In planning a social media survey, one of the primary considerations is whether the target population accesses that platform. There are likely to be certain demographic groups excluded from any social media platform, and if these groups are to be targeted, a supplementary form of surveying is needed. In fact, a 2016 study conducted by Sensis found that in Australia, 60% of internet users aged 65 and over, and 37% of internet users aged between 50 and 64 don’t use social media (www.sensis.com.au/socialmediareport). Comparing this with the fact that 90% of internet users aged between 18 and 19 do use social media, age must be a consideration when social media is chosen as a survey tool.
Self-volunteer surveys are in abundance on social media. The limitations of such surveys, delivered in any form, are well understood. The views of the people most likely to respond to a self-volunteer survey rarely represent the views of the entire population. Often those with strong views are more likely to respond.
More problematic are the algorithms and filters used by a social media platform to determine a user’s newsfeed and thus affect who receives the survey invitation. These algorithms can create a very real bias to survey results and it may be impossible to estimate this bias.
While such issues must be accounted for, social media certainly represents an efficient means for conducting a survey in certain contexts. What is important is ensuring the survey objectives are met and any biases are carefully addressed when interpreting and reporting the results.
To make the most of your investment in a survey project, ensure you get the right advice early on. Data Analysis Australia has over 25 years’ experience in designing, conducting and analysing surveys. We pride ourselves in our ability to maximise return on investment by designing cost effective methodologies to meet survey objectives, conducting informative analyses, and presenting the results in a meaningful format.