15 June 2018

One of the most well-known approaches for study and development of psychosocial working conditions is the job demand-control -model (Karasek 1979). In those days it was nearly impossible to image this modern age, when plenty of data concerning work is available in different digital sources.

Between old and new

Over decades data has been gathered through surveys. This means tens if not hundreds of new and modified scales for participants to answer, and traditions of analyzing and interpreting results. Instead of surveys, work data is accumulated in IT systems without our guidance and models. What kind of data this digital work data really is and how does it differ from survey data? Can we bring the old models about psychosocial conditions to bear this new data source? How to analyze and interpret digital work data and how to take advantage of this new opportunity? Scholars in different disciplines push on to discover solutions for these questions.

Work as evolving process

Working life is under continuous change and we can’t get valid knowledge about dynamic phenomenon using surveys once a year. Dynamic phenomenon needs more frequent and longitudinal approach for data gathering. When working using IT, digital footprint creates event data. Events can be characterized by time, place, actor, tools, network and other contextual factors. This event data, work data, is qualitative, situation specific and process-like data – not quantitative. Analysis of work data is in need of new methods created by data scientists.

Let’s return to Karasek’s model. If he is now following events in work data, he may recognize the pulse of work: Work does not flow as planned from one event to another. In a short time period there may occur plenty of different unplanned things and in some other time period fewer. It is possible to identify subgroups of workers from work data, with load peaks one after another, and subgroups with continuously evolving and varying situations.

Focus on intensity

Workload and time pressure have been since 1970s the core demands that affect well-being. Increasing control over demands has been the core enhancing element when designing work. In the future, under continuous change and pace, it may not be possible to develop any dimensions of control. Moreover, employees may be left alone as controllers of their time, yet be tied to and connected with partners’ and customers’ schedules. So, increasing control is not the solution any more. Should we focus straight to the demands? Intensity may be the new concept for the new demands described above. On the other hand, opportunity to use data, which is accumulated without researcher’s preconceptions, enables discovery of knowledge through grounded theory and data-driven approach.

New concepts for pattern recognition

To analyze digital work data, we need new concepts for feature selection and pattern recognition. The former refers preparation of raw data for creating a set of characteristics that reflect the phenomenon in question. Using features, it is possible to identify patterns by applying different methods of machine learning and artificial intelligence. We may interpret patterns as profiles of work practices, which may have an association to e.g. sickness absence.


Silent signals from work data -project is in progress (2018-2019). The project is supported by the Finnish Work Environment Fund (project number 117317).

Data scientists Andreas Henelius (Aalto University) and Antti Ukkonen (University of Helsinki) have engaged in the planning of the project. The project is carried out by work- and organizational psychologist and project manager Tiina Kalliomäki-Levanto and data scientist Jussi Korpela. The participating organization is University of Applied science, where HR and IT experts have together developed their data repositories to improve teaching and learning.