Human nature and free will are the worst enemies of research. I found this to be particularly true recently during a groundbreaking study which, unintentionally, discovered the following: Drivers, when questioned about their driving habits, will invariably lie about their adherence to the law.
To be fair – we don’t necessarily think the study participants intended to lie about their driving habits; instead, much the same way that an angler recalls their catch to be bigger than it actually was, these drivers tended to recall a better self-image of their doppelgangers behind the wheel.
But first, some context: This past winter, researchers from Escalent had the chance to partner with three very bright Michigan State University (MSU) students who were working towards their M.S. in Business Analytics. As part of their Capstone project, these students were required to work on a real-world, big data project prior to graduation.
We saw a dual benefit to the students’ partnership – we were working on a project that required the analysis of a large amount of vehicle telematics data from both passenger and commercial vehicles, and we needed the help! But for these students – who will soon be on the job market – it was a valuable way to help them understand a mega-trend in employment.
According to the latest GreenBook Research Industry Trend (GRIT) report, companies are on the hunt for new skill sets to meet the demands of a rapidly changing marketplace. Further, the report found that data analysts and data scientists are in high demand. The MSU program is dedicated to preparing young talent for the future of connected devices, and for the future of mobility by partnering with companies who are also preparing for a new generation of consumer engagement and business models.
With our team assembled, we set out to understand the relationship between self-reported driving behavior collected via an online survey, and observed driving behavior collected via an onboard telematics device.
As researchers, we know there are limitations associated with every research project, and there is an array of reasons why participants don’t, or can’t, provide accurate responses. Of course, we design research projects to minimize limitations, but it is impossible to account for all factors that may influence research findings. As such, telematics offers a rare opportunity, through technology, to better understand the relationship between self-reported and directly measured behavioral data.
Through the comparison of telematics data collected from 130 passenger vehicles against self-reported data collected from an online survey, the researchers and MSU students learned something surprising – the actual driving patterns of more than half (55 percent) of participants did not match their self-reported tendencies.
While we expected a degree of mismatch between the telematics and self-reported data, we, honestly, didn’t expected half the sample to be wrong. We also observed that men were more likely to misreport – both in stating they had an aggressive driving style but actually were passive, or self-reporting they were passive when they actually were aggressive.
We don’t believe our participants intentionally lied. Instead, we believe that a case of “social desirability” bias – the tendency for people to over-report “good” behavior and under-report “bad” behavior – was at work here. Another explanation may be something called “compromise effect.” This is a result of the tendency to choose the middle option, rather than options on the extremes, when presented with choices. This also can happen when options are not clearly stated, or are vague. Though we believed we had provided three clearly-written descriptions of driving styles to choose from, and while we put considerable effort into crafting these descriptions, we perhaps didn’t entirely hit the mark.
So what does this mean for survey research? While using data from a connected device – in our case, a vehicle – may not be possible for every research project, as an industry, we need to take advantage of enabling technologies that will allow us to better understand the extent of the gap between intended and actual behavior, where feasible or appropriate. I do not believe we will be able to forgo engaging with consumers to solely rely on passive data. But, we certainly can complement self-reported feedback with observed behavior for more confident business decisions.
If we at Escalent can be of assistance to you in conducting this type of analysis, please reach out to me.