“One of the major challenges we face in online primary data collection methods is data reliability and preserving the integrity of the data collected. So we continuously think about how to ensure the reliability of data,” Sushma Vasudevan, VP of Data Science and Analytics, said.
PureScore™ is a machine learning-driven model that is designed to evaluate individual respondent level quality on each and every transaction or interaction that a user has on the PureSpectrum platform over time. The model-based approach takes into account three things: the profile of the respondent, their past behavior, and their current behavior on the platform. The system will then assign them a score from 0 to 10. Respondents with a failing PureScore™are blocked from participating in surveys while those with a passing PureScore™ are allowed to participate in surveys on our platform.
“As PureSpectrum grew one of the things we saw was the kind of data we were generating. We reached a point where we generated so much data that we had to not only scale up our way of computing but also the way of looking at respondent behavior on our platform” Vasudevan said.
Any inconsistencies or negative respondent behaviors are immediately flagged and penalized. This consistent monitoring ensures the reliability and integrity of the data collected on the PureSpectrum platform.
The Morphing Nature of Survey Fraud
“A lot of times within market research, people will use the analogy of a virus for data quality fraud and how it will mutate and become resistant to the techniques or treatments that have been developed against it,” Mark Menig, PureSpectrum Chief Product Officer, said.
The morphing nature of fraud is evident in market research. Fraudsters adapt and stay ahead of data quality initiatives, continuing to fool researchers and collect incentives. In order to stay ahead, quality-ensuring methods need to be constantly evaluated and improved. PureSpectrum has worked to address these issues and presented PureScore in early 2020 to lead the quality initiative in the market research industry.
“Just because you think you have stopped them, they are going to look for a way around what you’ve done. That’s the tug of war, and ongoing nature of it that requires the attention and the innovation to keep yourself protected” Menig said.
PureScore™, a respondent-level scoring system, is driven by advanced Machine Learning technology. The model processes millions of transactions in real-time, allowing it to learn ideal respondent behavior. As the PureSpectrum system continues to adapt to changing respondent and fraud trends, it allows researchers to meet the changing needs of the industry. Continued data validity and reliability remain essential to researchers.
The Importance of Data Validity and Reliability
PureSpectrum is committed to providing industry-leading data quality methodologies. Our team of experts believes in empowering PureSpectrum users with technology that ensures data reliability and can better inform their business insights. The PureSpectrum platform features top techniques built to combat modern quality challenges and keep up with the ever-evolving data quality landscape. By constantly advancing methods to catch and prevent potential fraud as well as increasing data validity and reliability, PureSpectrum sets the standard for data quality in the market research industry.
How to Ensure Reliability of Survey Data
Quality control in online surveys can be difficult if not impossible to ensure manually. However, this is one of the built-in benefits of using the PureSpectrum Marketplace. Once Marketplace suppliers are vetted and onboarded onto the platform, monthly performance evaluations are conducted to confirm that only high-quality sample is being delivered to surveys. Further, all respondent behavior on the platform, including but not limited to items like answer consistency and length of interview, etc, are monitored on an ongoing, real-time basis. All behavior affects a respondent’s PureScore™ to ensure optimum data reliability.
“At PureSpectrum, we don’t expect researchers to have to worry about data quality. We expect them to know it’s being addressed and know that by selecting to work with PureSpectrum that they are choosing a platform that has made a commitment to data quality” Menig said.