The experiment, which used an artificial simulation of an active panel, saw panellists provided with a tablet computer and asked to report their TV viewing, with their viewing also passively monitored by ACR and TVision’s Computer Vision platform.

The findings showed that when the television was on during the four-week study, the simulation found that passive and active metering approaches agreed on viewer counts only around 56% of the time.

In the simulation, 18 to 34-year-old viewers were as much as much as seven times more likely to under-report viewing than over-report when measured through an active meter panel.

Viewer fatigue was seen over time on active panels, with underreporting going from 23% in week nought to 36% in week four and over-reporting going from 19% to 12% over the same time period.

Error rates were much higher for older demographics, though over and under reporting by the group generally offset each other.

CIMM and TVision said that non-compliance on an active panel could be the result of panellist errors or technical faults, but the data could also represent actual viewing behaviour.

As a result, measurement providers need to decide how best to assign viewing, given the composition of the household – introducing more complexity and potential for error.

Jon Watts, managing director at CIMM, said: “The study helps to illustrate some of the issues that can arise, as vendors look to correct for non-compliance.

“The impact and effectiveness of these corrections over time may be difficult to determine, unless some kind of secondary measurement check is introduced on an ongoing basis to determine whether the corrections made are accurate.”

Yan Liu, chief executive at TVision, said: “When we consider all the opportunities for potential error in an active panel, and the additional cost burdens associated with effectively managing and reducing that potential for error, we can see that active panels present a significant cost burden for the industry.

“In contrast, passive meters eliminate non-compliance issues and require less intrusive interactions between the panel operations teams and the panellists. This results in greater accuracy, lower panel churn rates and lower operational costs.”