Ensuring Respondent Quality for Researchers
The data quality landscape is always changing, and fraudsters are continuing to find new ways to trick survey systems. To continually gain and maintain high-quality respondents means keeping up with these changes and innovating for the future. Researchers expect advances in technology from their sample partners, assuming all data is consistently reliable, of high quality, and efficient.
On the PureSpectrum platform, before a respondent even enters a client survey, suppliers are vetted and only the best are chosen to be included in the PureSpectrum marketplace. While interacting with the platform, respondent quality is ensured through:
Preventing Duplicate Respondent Entry
This makes sure that all traffic coming into the survey, regardless of panel source, is unique. We track this through a unique PSID tied specifically to one respondent across all of their panel subscriptions. Learn more about PSIDs and other PureSpectrum terms here.
Scanning for Survey Fraudsters
This includes link manipulation, hash failures, IP validation, Geo IP validation, device fingerprinting, and more.
Tracking Good Behavior in Surveys
How respondents behave in a survey, including LOI, red herring questions, answer consistency, etc.
Do they commit fraud while in the survey? How long do they take to complete the survey? Do they complete the survey or term out? Why?
Improving Respondent Quality with PureScore™
Other data strategies measure quality at the collective panel level, making it difficult to validate individual respondents. Poor quality answers could get into surveys on a notoriously strong panel, while valuable responses could be blocked by other panels. This prevents the complete blockage of low-quality respondents and limits feasibility.
Beyond traditional methods monitoring for fraud, we use advanced machine learning techniques to evaluate data quality and validate respondents on an individual level. Machine Learning enables us to process every transaction that comes through our platform, identify and predict patterns, as well as make optimized decisions, all without human interaction. Machine Learning is defined as “a field of study that gives computers the capacity to learn, without being explicitly programmed” by Arthur Samuel in 1959, who coined the term.
All of our respondent quality measures come together in PureScore™, a comprehensive, advanced Machine Learning driven, scoring system designed to measure individual survey respondent quality on a scale of 0-10. The model finds patterns in the types of respondent profiles and respondent behaviors. The more that a respondent deviates from the ideal, the lower their PureScore.
Any respondent with a PureScore of 5 and under is blocked from entering the survey, but can still redeem their score by providing valid responses on the PureSpectrum screener.
PureScore is the industry’s first respondent scoring system, allowing researchers further insight into the quality of their data on a transaction-by-transaction basis. Platform users can better trust their data, include only high-quality respondents and reduce costly data cleaning at the end of fielding. PureScore is the next evolution of modern data quality methods.
Since PureScore improves the quality of data that enters client surveys, it also drives down data cleaning requirements and reconciliation rates. Data reconciliation rates are typically hingent on the various methods and policies of the research, but PureScore can still improve upon these costly processes. After just the first month of implementation, our clients saw a 15% decrease in reconciliations and an increase in respondent quality.
Interested in learning more about how PureSpectrum can improve your respondent data and survey quality? Reach out to us today to schedule a demo.