Co-designing conversations — workshops and analysis of qualitative data.
I am blogging my MRES. This may be incomplete, poorly cited (at this stage). But forms part of a writing discipline attached to design research. This post shows the evolution of a co-design workshop from a joint activity to a game.
Spring-boarding from the semi-structured interviews. I developed a workshop looking at clinical conversations and how they might be improved. By focusing on theses exchanges, I wanted to research through design how, on a micro-level, the different actors within transitional care (Patients, HCP’s and Parents) relate.
A service design analysis of transition may uncover aspects of the interaction between these groups, but only on the level of the service. Most likely, it would show channels of communication, issues with accessing services, identify various service touchpoints and reveal emotions attached to service use (source). However, this analysis may be too broad to uncover why specific exchanges between actors break down.
It may show where the service is broken, but not show why it’s poor. A service design approach identifies issues with communication (such as poor frequency, lack of actions attached to a message) but may not be detailed enough to show why these events occur? Is it the content of the message not just its frequency? Or was it delivered by a system the (now disengaged) patient feels doesn’t listen to them? How might we uncover these details if the analyses tools are too broad?
Using games in co-design workshops
The workshop, evolved from a desire to look at transitional care form the perspective of a person with diabetes. Originally, it was primarily a co-design mapping task (see figure 01), with the researcher facilitating. However, failure to recruit enough participants in time (see appendix — when research fails!) resulted in more time to evolve the workshop activities.
This evolution lead to a number of game prototypes being play-tested with volunteers from the Design Research Studio at the Royal College of art, and clinical staff from St Mary’s hospital Imperial College Trust (See Figure 02). This was using, primarily, low-fidelity cardboard components to build the game and develop compelling game mechanics, rules and play.
This resulted in ‘Consultation’, a ‘game’ to help expose some of the mechanisms involved in conversations. It is part game, part co-design research tool. It is a turn based card game where the patient choses a role (there are only three roles: Person with Diabetes (PWD), Parent and a HCP), and participants choose from a number of cards, which they play as the conversation progresses. These cards range from emotions, to actions, to tasks, enquiries and location. The goal is to reach an agreement (any agreement) about the starting topic (full rule-set and images in the appendix)
To focus the research, I wanted to look specifically at why conversations were sometimes poor, or why they were considered key collaborative encounters supporting good condition management? A game, with associated props and mechanics (cards, roles, turns, etc) is a scaffold to explore this. The participant can easily react against elements such as rules (“i didn’t get enough turns)’ or respond to perceived play outcomes ( “it was unfair” or “I liked how I could do that!”). These are concrete reactions to specific mechanisms with the game.
Participants were recruited as part of the the monthly transitional and young adult type 1 diabetes clinics at St Mary’s hospital Imperial College Trust. There were 4 sessions in total of which 4 there were for participants (n = 4). Figure 1 shows a breakdown of participants, and the variety of players and roles. In some cases I played a ‘role’ in other cases I was the facilitator.
Research risks with recruitment
Participants were invited after their clinic appointment. Clinic attendees were emailed a day prior to clinic, notifying them of my presence in an attempt to increase uptake. Research like this relies on participants having enough free time to participate.
There is a risk with recruiting people post clinic, even if they are forewarned. Clinical appointments may have been long and arduous. Participants may be fatigued and unwilling to engage with a game that mimics their previous clinical encounter. Especially if that encounter that had been difficult.
Inversely, there may be a risk in recruiting only fully engaged participants, whose responses may be limited to generally positive aspects of T1D care.
The game was set up in an empty consulting room, in close proximity to the transition and adult clinics. For each session, the game was played once, and a video of the session recorded. Each session took around 10 minutes to play the game. This duration of encounter was expected considering the research was done in clinic and time was expected to be an issue.
In some sessions, the researcher played the role of parent. In others,where there were two players, the researcher played the role of facilitator. In all cases, with a new game like this, the rules and game play still require additional refinement. Facilitation was always necessary.
Analysis and findings.
Analysis proved challenging as this novel technique forced the considering of alternative analysis techniques. There may be other research analysis tools that may be adapted for this process, that may prove more fruitful, but for the research I focused on breaking down the conversation to its constituent parts in order to map what was said against the mechanics used to express it.
Firstly, techniques were adapted from grounded theory. Primarily, the repeated playback of each session to improve understanding. After these viewings (the sessions were only 10 mins in length), I mapped key concepts and quotes to the visual record of the conversation: the played cards.
This not only allowed an analysis of what constituted a good conversation, but also allowed the comparison between good and bad conversations. However, a good conversation was based on the researcher assessing, post session, the depth and engagement of the participant(s). This ranged from very poor, poor, good, excellent. There may be issues with using the researchers personal assessment of the conversation.
Respondent: 19 year old male.
Cards Played: 8
PWD played PWD: 3 cards played
Researcher played Clinician: 5 cards played
Conversation rating: Good.
The participant is highly engaged with their condition and its management. Conversation focused on one aspect, managing BG levels at work. It was a general discussion, with no specific management techniques, with the agreed action to focus on. Both players explicitly stated their confidence in managing the BG levels at work.
The respondent found the exchange familiar, in that it mirrored existing clinical conversations. How, the respondent described his typical clinical encounter as being “around 5 minutes after a 1 hour wait”.
“It’s important that the conversation happens but it doesn’t feel pressured” — PV0001
Respondent: 16 year old female.
Cards Played: 6
PWD played PWD: 2 cards played
A clinician played clinician: 4 cards played
The game was played with two participants, a clinical psychologist and a person with diabetes. The participant was 16 and from the position of the researcher, seemed nervous and unsure how to act. This meant the researcher played a greater facilitator role.
The participant was asked to compare this interchange with her clinic appointment, which happened 5 mins previous. She pointed out that in clinic, they didn’t speak about ‘talking on the phone’ even though for her, this was a really important channel to get advice clinical advice on.
Respondent: 20 year old female.
Cards Played: 6
PWD played PWD: 2 cards played
Researcher played Parent: 4 cards played
The patient engaged fully with the interaction and together, explored parental governance of diabetes in their children. This was a rich exchange, in part I believe, due to the enthusiasm of the participant, but also because the issue explored seemed particularly important.
The initial conversation, focused on emotions attached to parental involvement in care. Of particular concern for the PWD was the expectation of ‘perfect’ condition management from their parents, and the anger and questioning she expected when parents found out about occasions of suboptimal control. She felt that the expectation of ‘good control’ was a high burden to carry, especially with a condition like diabetes, which is so difficult to get right all the time.
In her words “If you don’t have diabetes, you don’t know how hard it is. People don’t realise that sometimes we’ve had such a rough day, we’re scared to go to bed. We can’t tell our parents, because they will worry more and add to our stress”
The mapping shows the back and forth of the conversation. The participant detailed her frustrations in overly strong parental questioning by playing the ‘anger card’. As the “parent’ I expressed the theme that came out from the SSI’s that stepping back from having so much parental control was difficult.
This lead to the mature expression from the participant that they to feel they can make mistakes in management without the added stress of parental judgments on their perceived failings.
This was resolved with an action card that specified an agreed number of times per week that the parent can enquire about the participants condition management.
“I need this game at home! we do come to an agreement…but its after a lot of arguing!”
“Yes, the “let’s do this” [card]..this give us options and the freedom to pick and we’d come to an agreement. More healthier than a back and forth argument”
Respondents: 16 year old female and her Father (age unknown).
Cards Played: 6
PWD played PWD: 2 cards played
Parent played parent: 4 cards plated
The conversation was very difficult to start but eventually settled on a conversation around the patient’s overall management of diabetes. Generally they were happy with all aspects of their care (which is surprising) and wanted to express that sentiment. I suspect their expectation was that the workshop was a simple feedback session.
The researcher played the role of ‘facilitator’ which I believe was problematic as the researcher became the focal point of discussion. The participants spoke to the researcher and not to each other. To facilitate the session, the researcher progressed the discussion by playing cards. In effect this moved the exchange to be closer to a semi-structured interview about condition management.
Interestingly, when concluding, the parent suggested what they should do is an ‘emergency fire drill’, in order to train the household in what to do if there is an emergency related to diabetes.
A number of findings emerged from analysis of the mapping of the conversations.
The only ‘excellent’ conversation had the characteristic of having a large number of emotion cards played at the beginning of the exchange. This suggests that when tackling emotionally charged issues, it’s easy to shift into poor conversational behaviours around point proving and defending a position. Feedback from the participant suggested that this was the case when dealing with her parents and the mechanism of playing action cards, helped break any emotional impasse that may occur.
Furthermore, the playing of action cards with the goal of reaching an agreement, seemed to focus the conversation towards actionable behaviors.
Participants under 18 (PV002 and PV0004) seemed unsuited to playing the game. Even with such a small sample size, there is a marked increase in engagement for people who were 18. Players under 18 seemed hesitant to select a topic, with both participants needing help from the other player. In both cases the partner (facilitated by the researcher) played “Continue” cards early in the exchange to try to move the conversation along.
Good conversations seemed to emerge as a result of emotion followed by action. In fact the only excellent conversation occurred when the patient played the agreement card (PV0003). This seems to mirror findings from the semi-structured interviews that suggested that ‘the best clinics are ones where they work with you”. This collaborative exchange seems to reflect a good mix of the emotional cost of dealing with a long term condition, combined with concrete actions to mitigate that burden.
Analysis of the game as a research tool
Designing a game as a tool for design research had many advantages. Its valuable to assess aspects of this research technique outside of any findings, in order to improve or adapt the technique.
A game can be a conversational map of what elements were discussed, which the players can then use as the basis for discussion. For ‘Consultation”, the cards people play are also a tool in assessing of the encounter. The game can directly expose the conversational mechanic, which after the game is completed, can be used to compare to previous conversations. This comparison can show where things were better or worse, and perhaps describe why they were better.
More broadly, games can support shy or less confident participants. As they can be proxies for human behaviour (Schell or Koster), people may behave (or feel they can — how many times have you bankrupted your grandmother in monopoly?) in ways they wouldn’t normally in reality. When generating insights into improving, transitional care, frankness from subjects in research is highly prized. Especially those who are less engaged. However, the fact that younger players (under 18) had trouble with the game shows there is more work to be done in this area to make it accessible.
Games also provide props for people to behave in different ways. Rules and mechanics gives players clear courses of action. Games can therefore be a powerful tool in helping people overcome barriers or mental blocks with regards to a particular topic, idea or opponent. As the researcher when playing the role of parent, I could facilitate the conversation towards a shared action by playing action cards. This may seem a forced exchange but it shows the value of props in providing a structured to have conversations around. By opening up new actions or new routes, highly emotional, repetitive exchanges that achieve little may be avoided. This moves the conversation to a point of shared understanding by creating a shared goal.
Furthermore, the most novel aspect of ‘Consultation’ is for people to play different roles. Within conversations about T1D, a person with diabetes, almost always plays the role of a person with diabetes. This means most actors rarely understand the role of diabetes management form the perspective of the different actors. Playing ‘Consultation’ makes it easy to usurp those established roles. What does this mean for a person engaging with clinical care if they better understand the factors that influence how care is delivered? The impatient clinician, the overbearing parent, the dis-engaged teen may be then seen in a new light.