There are two key points to note about our sample structure.
1. WE REPRESENT 16-64 YEAR-OLDS
We don’t interview anyone aged 15 or under because parental consent would be needed, and because a parent or guardian would technically need to observe them filling out any survey. This makes it slow and expensive to conduct fieldwork, and it’s possible that some minors would not answer all questions accurately if being observed by an adult.
We don’t interview people aged 65+ because it’s very challenging to find enough people in this age bracket in certain markets – particularly emerging markets where internet penetration rates are low and online populations are therefore dominated by younger age groups.
As we want our data to be representative and harmonized across markets, we therefore set an upper age limit. A long-term ambition is to remove this limit and begin surveying older internet users, but this is unlikely to become a reality in the short- or medium-term future.
2. WE REPRESENT ONLINE POPULATIONS
Because internet penetration rates can vary significantly between countries (from a high of 90%+ in parts of Europe and North America to lows of around 20% in parts of APAC), the nature of our samples is impacted accordingly.
Where a market has a high internet penetration rate, its online population will be relatively similar to its total population and hence we will see good representation across all age, gender and education breaks. This is typically the case in North America, much of Europe and places in APAC such as Japan and Australia.
Where a market has a medium to low internet penetration, its online population can be very different to its total population; broadly speaking, the lower the country’s overall internet penetration rate, the more likely it is that its internet users will be young, urban, affluent and educated. In some Middle Eastern, African and Asian countries (e.g. India, Indonesia), we would also expect a gender-based skew towards males.
To ensure that our research is representative of a country’s online population aged 16-64, we set quotas on age, gender and education. To set these, we conduct thorough research across a range of international and national sources. At a global level, these include the World Bank, the ITU, the International Labour Organization, the CIA Factbook and the US Bureau of Labor Statistics; within individual markets, we are typically taking data from national statistics sources, government departments, Eurostat or other credible and robust third-party sources.
In an ideal world, most researchers would probably decide to set quotas on income too. However, details of income can be a sensitive matter for many people, and some will always prefer not to disclose this information. As a result, it is not best practice to insist that people share their income, which therefore prevents us from setting wealth or income-based quotas.
As a proxy for this, we set quotas on educational attainment – distinguishing between those who have achieved primary, secondary or tertiary level education. An additional benefit of this is that educational achievement is a relatively stable metric, whereas household or personal income can be subject to considerable short-term changes and fluctuations.