Smart Automation for Higher Data Quality

In today’s world, the reduction in time allowed between end of field, production of results and presentation to stakeholders, is often so short that any data quality issues quickly become serious business issues. Once stakeholders start to lose faith in research, all can be lost. It is no surprise that the sample supplier is judged to be at fault, and buyers go looking for alternative sample sources. Yet, creating feedback loops between sample suppliers and buyers with both parties working closely together to combat these “survey cheaters” presents a better approach to subvert this problem in the future.

Sample quality is clearly identified as a problem that has affected many people – and some of them quite often. The real problem is one of identification: not “how fast can you spot you have a problem?” – that is sometimes blindingly obvious – but “how fast can you identify the truly poor records?” and “how quickly can they be replaced?”

Sometimes a cheater cohort is easily identifiable: they make little effort to disguise their cheating and give themselves away. These are the cases you can find through classic data quality traps and checks. These checks however are becoming increasingly less effective against what we might call the “smart cheater.” They know what we are looking at, and they make efforts to provide us what we need: good looking open-ends (often cut-and-paste from the internet), surveys done at a reasonable speed (by waiting at the end of the survey to click “finished”), little or no straight lining, etc. They cannot be seen by the naked eye, but they can be seen by an AI working with all its machine learning to compare the results of the cheaters with the results of the majority – for that is the one thing the cheater does not know: what is the “right” answer, and what is its distribution. And of course, the AI also finds the “lazy cheater.”

Because they work in real-time, automated solutions such as QualityScore™ from Dynata’s Imperium, prevent bad cases from ever becoming a completed survey. There is therefore no knock-on effect on timing, no missed deadlines, fewer cases of planned sample sizes not being achieved, and much reduced chances of poor business decisions being made.

This is the future. This is the solution.