For a project involving diabetes management, we were asked to provide support for the logging of participant mood/emotions. While some apps just take a few smilies to let the user encode his/her mood state, we decided to investigate some literature from psychology in order to find a new sweet spot between theoretical underpinning and user convenience. In short: initially we only found too long surveys with a too high bias towards psychological disorders (like depression), eventually, we found questionnaires involving 10-20 questions. Then, we realised that for the sake of experience sampling, we did not have to ask all these 20 questions every time.
In the following, we first present the current GameBus prototype for mood logging and then we link it back to the literature survey that we had performed. In any case, we also like to thank Pascale Le Blanc, our psychological partner in the GOAL project, for sparring with us on this matter.
Demo and Evaluation
For an interactive demo, please visit https://app.gamebus.eu/nav/activities/create/eyJzY2hlbWUiOnsiZGVmYXVsdCI6eyJpZCI6MTA2MiwidHJhbnNsYXRpb25LZXkiOiJMT0dfTU9PRCIsImltYWdlIjoiaHR0cHM6Ly9hcGkuZ2FtZWJ1cy5ldS92Mi91cGxvYWRzL3B1YmxpYy9NVFUzT1RFM05EUXhORFkzTTFwbU9GUnBhVXBqLnBuZyJ9LCJhbGxvd2VkIjpbXX0sInJ1bGUiOm51bGwsImNvbmRpdGlvbnMiOm51bGx9 (note: you will need a GameBus account). In that demo, you notice that we collect two main items:
- An emotional state that the user wants to register, and
- A classification of how strong the user feels that emotional state.
We allow the user to register such moods multiple times, so if a user has a combined feeling then GameBus enables to convey that. For example, a user can first declare to feel “excited” with a strength of “quite a bit”. Then, the user can declare to also feel “scared”with a strength of “a little bit”. When a user only enters these two emotions that day, we assume that he/she considers his/her other emotions irrelevant.
Video Demo
Here is our own rough evaluation of this approach:
- strength: unlike classic questionnaires, we do not force the user to go through all 10-20-60 items that psychological scholars find interesting. This lower burden on the user may free up energy/motivation for expressing his/her emotions more frequently (as it takes less time per registration session).
- weakness: some users may feel systematically avoid registering an emotion, even though they feel it. That may be triggered by many processes related to guilt/shame/… and it may bias further data analyses.
Our intention is to mitigate the above mentioned threat/weakness by creating a safe environment for the data entry, by providing adequate privacy and security measures.
Relation to the Literature
Initially, we had been exploring depression-related literature, covering for example the Self-Rating Depression Scale (SDS) and the Clyde Mood Scale. Such work seemed promising as it has already been very thoroughly studied since 1965. We eventually did not push through, as the related surveys appeared to be designed more as intake/discharge surveys while we were looking for questions to be asked more frequently (in line with the to the ESM/EMA approach). We also felt that various questions related to these scales were potentially introducing a measurement bias, as the questions may negatively bias towards negative emotions associated with depression.
Second, we have explored the research line on the Positive and Negative Affect Schedule (PANAS). We were very happy to discover that research line, as it explicitly covers positive and negative emotions. We were also happy to find results from many efforts to reduce the length of the related questionnaires. Still, even the shortest form (PANAS-X) still had 10 questions, which conflicted with our study design in which we would sample frequently and cover not just emotions but also food intake, etc. Therefore, eventually, we decided to first move back to the more elaborate “full PANAS” (covering a taxonomy of 20 emotions) and then let the user decide which specific emotion to convey information about. We also did preserve the scale per emotion from PANAS (i.e., we still have the original 5-point likert scale with exactly the same labels from English and Dutch studies based on PANAS). Throughout our literature survey, we also came across other interesting questionnaires and scales. In particular, we were charmed by the work of a 2016 student from UTwente, covering the Brief Resilience Scale, Dutch Version (BRSnl)). We did not pick that instrument for use in GameBus since just like SDS it seemed more suitable to one-time longer-term self-assessments than high-frequency/momentary self-assessments.