Prize draws and 'over-sampling': How researchers are hunting media's missing audiences
Some people in the UK are ‘hard-to-reach’ and risk being under-represented by media owners and advertisers. What are research companies doing about it? Ella Sagar reports.
Sometimes advertisers and their media agencies have some very specific audience requirements.
Take this example:
“C1 and C2DE women aged 25- 54 that had young male relatives aged 11-19 years old and they all had to live in the [London] boroughs where knife crime was most prevalent.”
This was the “really niche sample” in 11 designated London boroughs that the Metropolitan Police and its media agency Wavemaker had to reach with a hard-hitting anti-knife crime campaign in 2021.
Not only is identifying people that match this criteria an involved project, but what happens when many of these people don’t want to engage with researchers who come to their door?
Wavemaker’s insight team created an online tracker to survey 250 women in each wave of the campaign that met that criteria.
Sometimes when brands need to reach a very small sample of the population for campaigns they use specialist research agencies like Alligator, which Wavemaker used in the above instance, as discussed at the launch event for the Adwanted Media Research Awards (AMRAs).
However, there is a wider problem in recruiting and retaining specific demographics for surveys which contribute to nationally representative media research and audience measurement for joint industry currencies (JICs).
This means certain segments are designated as “hard-to-reach”, which could refer to: people of whom there is a low percentage or incidence in the population, people of whom a low percentage will acknowledge they belong to a particular group (ie. it could be a sensitive topic), people who are very unevenly distributed and difficult to find, and/or people who researchers are worried are being under-represented because of sampling issues.
Belinda Beeftink, research director at IPA, told The Media Leader: Measuring audiences is “not a simple thing” and that increasingly the media and advertising industry relies on a combination of survey and first-party data to get scale and granularity in overall measurement.
“It is becoming harder and harder to get certain cohorts to take part in surveys — specifically young people and more specifically young men. This is not just a UK problem but an international one and one which many of the research companies are grappling with right now.”
She added: “Another by-product of audience fragmentation is the need to ensure that any data collected is representative. It is still not good enough to rely solely on online surveys when there is still a proportion of the population who are not online users or who are very light online users. If you rely on people who are heavy internet users, you run the risk of skewing the data collected around media consumption.”
In addition to young men, Beeftink said ethnic groups are “always hard to recruit “and older people tended to be easier to involve in a survey because they have more time.
Some other reasons why certain groups might be hard to reach could be that they are unlikely to be at home for a face-to-face interview, and if online is tried, they could be less likely to look at the content unless it is relevant to them, or they can be less interested in taking part in research.
How do you find the ‘hard-to-reach’ audiences?
Beeftink said there were many ways to tackle the problem with recruiting certain groups for surveys which could include, but are not limited to, cloning, ascription, and weighting.
Audience measurement companies Barb and Rajar recognised this challenge of recruiting and retaining specific demographics for their respective panel and diary-based surveys.
Caroline Baxter, research operations director, Barb said its representative panel is growing to 7,000 households, or approximately 16,000 people by next year.
She said the make-up of the panel is determined by Barb’s Establishment Survey, a continuous survey of 53,000 household interviews annually which are normally conducted face-to-face.
The Establishment Survey draws on UK Government census data to set targets for the number of households from different demographic groups that should be on the panel.
Baxter explained: “Certain groups are harder to recruit to and retain on the panel, such as ethnic minority households and younger, pre-family households. We, and our panel recruitment contractor, Ipsos, are continually looking at how we reach these groups and investing in techniques (such as disproportionate sampling) and materials to improve their recruitment and retention, to ensure we are meeting the targets set by the Establishment Survey.”
Disproportionate sampling is where researchers focus a sample in specific areas where they know certain groups are and/or increase their sample in those areas to enhance the opportunity to find them.
Other solutions include targeted materials aimed at those hard-to-reach groups, for instance translated into different languages, and techniques like matching interviewer profiles to those groups, such as interviewers that have specific language skills.
Neil Farrer, head of audience measurement at Ipsos UK, added: “Panel recruitment is a core foundation for Ipsos. Building a representative sample of individuals across a range of demographic and geographic groups to take part is key. There can be challenges around recruiting certain demographics, however finding new and innovative ways of reaching out to and engaging with these people in a way that they find attractive is important.
“Often this means that we need to be very strategic about reaching and recruiting niche groups, or that we need to assess the compensation offered for taking part. Ipsos has a long pedigree of successfully building representative panels across audience measurement and public affairs, primarily because we understand our panellists and put them first.”
A spokesperson for Rajar said young people were harder to recruit, so it compensates by “over-sampling”. This could mean “boost assignments” where interviewers only recruit 15- 24s and male 25- 34s.
Rajar’s universe includes all individuals aged 10 and over living in private households in the United Kingdom and recruits face-to-face and asks the participants to take part in the survey that week.
It has quotas for each area it surveys which are based on the demographic profile of the area. It has specific targets against 15- 24, 25- 34, 35- 64, 65+, working status, household size and ethnic origin.
There are also additional incentives like prize draws for children 10- 14 years old, 15- 24s and males 25- 34s as well as a monthly prize draw for all respondents who return their diaries.
Rajar has also invested in a targeted diary panel of around 700 to boost the representation of young people and selected ethnic minority groups. This means Asian “boost assignments” have been paused.