Protect the quality of your survey data from speedsters, cheaters and bots by implementing smart trap questions.

Posted on: January 19, 2021

Gathering high quality responses from online surveys is a fundamental premise of building customer insights and making sound business decisions. Achieving this premise, however, can be a major challenge. This is especially true when respondents are sourced from broad general audience panels and whose identity can’t be verified for privacy reasons. While collecting data from a reputable sample provider and using good survey design techniques go a long way towards ensuring data quality, it is often not enough. Inevitably, some survey respondents will set out to complete survey questions as fast as they can, particularly when a monetary incentive is involved. Not filtering out unqualified, inattentive or fraudulent respondents will at the minimum add

The curious case of failed electoral polls: Four take-aways for political pollsters from a customer insights researcher

Posted on: November 11, 2020

The growing success of new commercial online survey techniques and technologies may offer a roadmap for improving polling performance  By the time all the votes are counted, the ultimate margin of error may not be as huge as it appeared in the morning after the 2020 elections. However, it is already fairly certain that pollsters continue to struggle mightily with predicting election results despite conducing countless of polls. This struggle is particularly apparent when we look beyond national averages and start dissecting the errors on a state and district level. I am sure that there

Matrix reloaded: impact of matrix question length and attribute complexity

Posted on: October 16, 2020

Katherine Stolin [1], Charles Kennedy [2], Rastislav Ivanic, PhD [3] Abstract  In our earlier paper on the topic of matrix questions, we explored findings from two parallel studies using alternative formatting of a matrix question. In that data, we found evidence that the traditional table format of a matrix question resulted in a small but statistically significant drift of mean scores toward the middle of the answer scale. To confirm this finding, we have conducted a