Study: Panels key in big data audience measurement
December 13, 2023
While big data streams from smart TVs and set-top boxes allow for granular and near real-time measurement of viewing audiences, the latest ARF DASH TV Universe study illustrates the extent to which panels are necessary for calibration to account for coverage gaps and biases inherent in big data sets.
DASH is a granular enumeration study of consumer behaviour in TV and digital media that is enabling measurement, media and identity companies to calibrate big data sets for greater accuracy and representativeness in audience measurement. The ARF, in conjunction with NORC at the University of Chicago, surveys a national probability sample of more than 10,000 US households annually in two waves. The Spring 2023 wave gathered responses from 6,219, or roughly 60 per cent of the annual target panel.
“The biases and coverage gaps in ACR and return-path viewership data have long been known, which is why the ANA’s Cross-Media Measurement initiative and standards from both the US JIC and the MRC call for panels to be used to calibrate these data sets,” says Paul Donato, Chief Research Officer at ARF. “The latest data from the ARF’s DASH study highlight why calibration with panels is needed and how our universe estimates can help providers enhance the accuracy and completeness of their measurement.”
The report uses the latest DASH data to highlight a variety of calibration cases, including these:
- Demographic assignment: Most companies use third-party data providers to assign demographics to households providing ACR and return-path viewership data, but the quality and coverage of those demographic profiles can vary widely, especially for young and more transient people. The precise household-level demographics and TV usage data in DASH can be used to assess the accuracy of third-party demographics and model the gaps. For example, the latest DASH data shows that the conditional probability of a household with a Disney+ subscription having a child is 43 per cent vs 28 per cent in the general population.
- Viewer assignment: In addition to providing comprehensive co-viewing measures, DASH can be used to estimate viewing in segments unmeasured and undermeasured by machine-level viewership data. For example, viewing on TV sets receiving signals by antenna (OTA) can be refined with an understanding of streaming in those households. The latest DASH data shows that OTA households are a third more likely than Pay TV households to stream during Primetime. Three-quarters of OTA households headed by 18-34 year olds streamed yesterday during Primetime, almost twice the rate of all Pay households and nearly on par with the rate of all broadband only (BBO) households.
- Householding: As more people young and old stream and watch linear TV on devices, householding the devices, and the viewing associated with them, becomes essential. The latest DASH data shows that 43 per cent of 18-34s watched TV on a smartphone yesterday, compared to 15 per cent of people 55+, but the gap narrows from 28 p.p. to just 9 p.p. for viewership on tablets (23 per cent for 18-34 and 14 per cent for 55+). DASH also captures use of connected devices on a TV set by TV set basis, enabling a precise understanding of who in each household is using which devices to watch what.
- Modes of reception: DASH breaks down Pay, OTA and BBO reception into their granular components, for more precise universe estimation and calibration. For example, the 33 per cent of all US households classified as BBO in the 2023 Spring wave includes 21 per cent with no live TV access and 11 per cent that receive live TV by virtual MVPD. The remaining 1 per cent is internet only.
- Duplication: Measurement companies may have tuning data from an MVPD and ACR data from an OEM for the same household. DASH can provide an estimate of the duplication. For example, the latest wave of DASH data shows that roughly 30 per cent of Comcast households had a Vizio television, and about 17 per cent of Vizio households had Comcast.