Peanut Labs attempts to crack the panel fraud problem
US-- Online sample supplier Peanut Labs is testing a system that claims to spot and block poor quality and fraudulent respondents from taking part in surveys.
The OptimusID system collects publicly available information about a respondent's computer and uses these data points to create a ‘fingerprint' unique to that person's PC.
Once a machine has been ‘fingerprinted', it can be identified every time a survey is taken – building up what Peanut Labs calls a “survey-specific behavioural history” for each respondent.
Using Optimus, the company says researchers will be able to identify and block any computers found to have taken too many surveys in a particular time period. It can uncover multiple email accounts used to take surveys on the same computer, as well as multiple panel accounts from multiple research firms – often a telltale sign of a professional respondent.
Other bad survey behaviour flagged by the system includes speeding through responses, straight-lining and satisficing, with persistent offenders sent to Peanut Labs' Respondent Rehab database.
“It has a three-strike rule,” chief operating officer Ali Moiz told Research. “If you do something wrong three times, you get flagged in this database where firms can identify you as a trouble maker and decide what to do with you.”
Moiz said the system used “an open-source data sharing model, so TNS, for example, can benefit from knowing if someone has a history of taking surveys badly for, say, Greenfield Online”.
But as any self-respecting celebrity knows, rehab is not forever. “Once you're in there you can get back out again,” said Moiz. “It's kind of like being grounded. After a month, your parents can let you out again to see if you've improved.”
The technology has been in development since early last year, while tests are now ongoing among 20 of the Honomichl top 50 market research firms, as well as end clients in the banking, gaming, packaged goods and entertainment industries.
“Despite the best efforts of the market research industry to fight the online data quality issue, bad respondents have persisted, skewing data quality in a severe manner that has forced us to develop this technology as a response,” said Moiz.
A similar system for spotting fraudulent respondents was developed by Düsseldorf-based Mo'web Research last year, however unlike Optimus it is not being licensed for use by other research agencies. Click here for our report on Mo'web's “guard dog”.
Author: Brian Tarran