Abstract
This paper explores strategies on how to best balance expanding survey length with the need for concise, relevant and engaging surveys, deployed in a device agnostic format. When designing a survey we, as an industry, are often seeking a balance between competing design challenges: clients have diverse and extensive objectives, survey participants have short attention spans and an ever increasing suite of connected devices to choose from. Survey participants are voting with their feet when surveys are not compatible with the device they want to use, whether that is the smart device in their pocket or laptop they are working on, and this is very real for online panels. We are seeing increased abandon rates, with the effects of extended fieldwork times, smaller pools of sample to draw from and the possibility of introducing bias into our data. Having spent much of 2015 working with clients to design more smart-device friendly surveys, Research Now has explored innovative ways to shorten survey length without compromising on the amount of material covered. Following on from work by Johnson et al. (2014), Research Now conducted a piece of primary research exploring survey modularisation as discussed in the current paper. The approach splits questionnaires into modules, with participants receiving only a specific module, a subset of the overall survey. It is expected that a long questionnaire can be split and – when applied appropriately, designed properly and implemented effectively – data can yield results comparable with a full non-modular survey. Building on previous industry work on this topic, and primary research conducted by Research Now, we discuss our methodology, the results and conclusions from this work, and explore opportunities to automate the approach. The overall goal of this study and resulting paper is to explore how adapting survey research in this way improves rather than complicates the lives of both researchers and research participants. If we are not able to shorten our surveys, then survey modularisation may prove to be our best hope for a complete, representative dataset and we need to ensure that this is achieved accurately, confidently and efficiently at scale.
Get full access to this article
View all access options for this article.