The Data Mixing Board is making the data gathered in the new survey accessible for more to create awareness and interest in how people meet and eat in and around the kitchen. It is an interactive tool where users can test their own hypotheses about life at home.
The Data Mixing Board is based on a survey in eleven cities around the world; Berlin, London, Moscow, Mumbai, New York, Paris, Shanghai, Stockholm, Copenhagen, Zürich and Sydney. The data was collected in two steps. First, we asked 8,527 people in Berlin, London, Moscow, Mumbai, New York, Paris, Shanghai and Stockholm about their everyday thoughts and habits in and around the kitchen. The data from these eight cities were collected between March 5th and 17th, and made part of the report insights. Between April 13th and May 12th we conducted the survey in three additional cities: Copenhagen, Zürich and Sydney. In July 2015 the data from the three new cities were implemented in the Data Mixing Board. All in all, 11,729 answers were collected, with respondents evenly distributed between cities. The median time to answer the survey was 13 minutes and 57 seconds.
Answers were collected among people living in cities with access to computer, in ages 18-60 years. The results from the survey are valid for this group; they are not to be generalized for the whole population of each city.
From the survey, key questions have been selected to be used as variables in the Data Mixing Board. There are topics described in the Life At Home Report that are not present in the Data Mixing Board. This is due to the necessity of creating a tool, which allows the user to do many things while at the same time not being confused by the complexity.
In the Data Mixing Board, it is possible to compare results between different income groups. To make this function meaningful, it is necessary to create spans that are somewhat comparable. This is challenging due to differences in currency, general income levels et cetera.
To make comparisons possible between cities, we have used the Big Mac Index as a benchmark. It is a good method since it mirrors the market conditions and the general purchasing power among the general public.
To produce the income span variable, we translated all currencies to USD and defined a general variable. This variable was then divided by the cost of a Big Mac in every country. The upper 25 percent and the lower 25 percent were defined as high and low income, respectively.
The survey focuses on different aspects of how people meet and eat in and around the kitchen, and shows both interesting differences and similarities between cities. One challenge when assessing differences between countries or demographic groups (e.g. men and women, age groups) tend to answer questions slightly different.
If we want to give a picture of all answers together, to find an “average” (in our case meaning the eleven cities combined), we need to take precaution of differences between groups. If one group is overrepresented in one city, biases could follow if that group also tends to answer questions in a deviant way.
To assess such group differences, we have analysed the material using a random forest technique, creating decision trees to find the groups or combination of groups that create the most significant differences for the scale questions assessing “life happiness”. We conclude that differences appear primarily between people with different occupations and living statuses.
When we control for these two demographic factors, we can look at the remaining differences between cities. Clustering the cities in two dimensions, based on how respondents answer the questions about happiness, we can produce a cluster map, showing which cities are closer related to each other in this sense, and which ones are further apart.
In the Data Mixing Board, it is possible to produce graphs illustrating all cities collapsed into one population. In this view, answers are weighted on occupation and living status to increase comparison.
The Data Mixing Board is an exploratory tool, meant to be used for testing hypotheses, while still being interesting and easy to use. It is not possible through the Data Mixing Board to validate statistical significant differences between groups or behaviours. The ultimate reason for this is that the data collection method has been online panels, which do not produce a random sample of responses (meaning that every person in a city would have the same possibility of participating). A random sample is a necessary criterion for drawing conclusions about statistical significance between groups.
IKEA does not take responsibility for any conclusions drawn through the use of the Data Mixing Board.