How does the data integration work?
In essence, data integration (also known as fusion) transfers audience estimates based on participants in one dataset to participants in another dataset so that there is a combined dataset that represents the data contained in both, as well as the relationships between variables from different datasets.
In order to do this, the fusion finds the best possible matches between participants in the two datasets in respect of their demographics and behaviour.
In the PAMCo fusion with Ipsos iris, exact matches are made for gender, age band (up to 65+) and Government Office Region (with some groupings). Other variables for which the best possible match is made are:
- Claimed digital behaviour
- Social Grade
- Working status
- Household Size
- Presence of Children
- Internet usage frequency
Each of the nearly one million Ipsos iris records appears once in the fused dataset and PAMCo survey participants in the digital part of the universe can appear multiple times but in a way that exactly preserves the original survey data.
Ipsos must align the Ipsos iris and PAMCo universes, as there are some differences (notably that Ipsos iris includes Northern Ireland). While the PAMCo survey data is completely preserved by the fusion process, the Ipsos iris data is slightly distorted but is almost always within +/-5% of the original GB part of the currency at the Total Digital level for monthly audiences.
The profile of the monthly digital audiences will match the Ipsos iris data in respect of age band, gender and grouped regions but will not match for other variables, as demographic data from the Ipsos iris data has been replaced with the demographic data of the fused PAMCo survey participants. As most variables present in the two datasets will not have been exactly matched at the most detailed level, differences in profile will be observed.