As part of the Core survey, GWI asks all respondents about their annual household income.
We state that they should think about their household income, rather than their individual personal income; this is to ensure we get a complete picture of their circumstances, which is not always the case when you ask about personal income only (especially for those people who might not be employed, but have a partner who is).
We also remind respondents that they should think about an annual, rather than monthly, figure. This is to ensure we capture comparable data across all markets, despite some countries being more accustomed to thinking about monthly earnings.
In each country, we give respondents a range of household income brackets to choose from; these are tailored to each market to reflect the reality that average household incomes differ between countries.
Until Q3 2016, this question had a free-text box where people entered an exact income figure. However, we decided to change to the current system of pre-defined brackets for two key reasons:
- Some respondents were reluctant to give an exact figure, which led to relatively high results for “Prefer not to say” or “Don’t know”. Once we changed to the new approach, we saw a notable increase in the number of people giving an answer.
- Frequently, respondents would shorten their answers – e.g. by writing 30 instead of 30,000; in many instances, it was clear how such an answer should be corrected, but there were other cases where we were having to discount the figure entirely. Similarly, in countries where currencies have high denominations and many people could have income amounts stretching to tens of millions, it was relatively cumbersome for them to enter their answer. This was exacerbated further by the tendency to shorten numbers, described above.
In each country, the income amounts are given in the respondent’s local currency. We are investigating adding a new element to PRO Platform where these figures are converted to USD and hope to have this in place by the end of 2018; we can’t ask respondents to answer in USD as few of them would know an accurate exchange rate, and asking them to make a conversion could lead to mistakes.
In some markets in Latin America and the Middle East, we are aware that respondents are still entering monthly figures despite us adding clarification text in 2016; we are investigating how to correct for this.
In each country, we use the results from the household income question to determine income segments – dividing people into the Lower Quartile / Bottom 25%, the Middle Quartile / Mid 50% or the Upper Quartile / Top 25%. We also break out the Top 10% (who still appear within the Top 25% too).
These segments exclude anyone who answered “Don’t know” or “Prefer not to say” in the income question; these individuals are instead included in the “Prefer not to say” segment. To allow a way to analyze these respondents alongside those who did give an income answer, we have developed a separate social grading classification which factors in much more in addition to income. You can find details of this here.
To assign respondents to an income quartile, we look at the spread of responses received in that country in that quarter’s wave of research. There are two key things to note about this:
- Respondents are assigned to a segment based on how their answer compares to everyone else we surveyed in their country only, and not against any respondents in any other markets.
Because average incomes can differ so much between countries, that means the thresholds of each segment are likely to be different from one country to the next; in a market like the US where average incomes are among the highest globally, the incomes of those in the top 25% are likely to be considerably greater than those in markets with low average incomes (e.g. African markets like Kenya and Nigeria).
This has an impact on how the data should be interpreted when analyzed at a global level. Rather than showing the top 25% of global respondents – which would cause a huge skew towards wealthy markets – it displays aggregated data for the top 25% from each country.
- As the segments are created after each wave of research and are based on the answers captured that quarter, the thresholds for each segment might be subject to wave-on-wave variation. In the US, for example, it might be that those with an income of less than 25,000 USD comprise the bottom 25% in one wave, whereas in the subsequent wave it comprises those with an income of less than 32,000 USD. In practice, it tends to be the same brackets that contribute to the same segments each quarter, but you can double check this by applying the income segments as audiences on the annual household income question.
As a result of asking people to select an income bracket rather than entering an exact amount, it is not currently possible to give a precise value for the threshold of each income segment. However, we can determine the income brackets that normally contribute to each segment by looking at the data from recent waves: