A influence on their total experience. This doesn’t only

A generative
the key for preference elicitation in large scale research

Author 38
Track 3: Eliciting patient preferences
through self-administrative context mapping

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now


Preference elicitation in large scale research (more than
50 participants) is a challenge. Many times, researchers return to conventional
user studies in the form of a questionnaire, but they often do not represent
preferences accurately. Participants have trouble with expressing their latent
needs and wishes and therefore give superficial answers, which will not lead to
better innovation. How can the outcome of these conventional user studies be
more accurate? This paper explores the field of mixed methods and proposes the
implementation of a generative introduction to preference questionnaires. It
mixes the flow of a generative session with the efficiency of a conventional
preference study. By implementing two main factors of a generative session, the
path of expression and making an artefact, participants will be supported in
the expression of their preferences. It leads to better preference election in
large scale research and thereby, more opportunities and better innovation.


Generative tools, conventional user
studies, preference elicitation, path of expression


Undergoing a surgery or staying in a
hospital for a long time, are life stages that a patient won’t forget that
easily. Events like these have a big impact on people’s lives. In order to make
these hospital admissions as comfortable as possible, it’s important to know
how patients experience those (Dekkers 2017). With the right information,
design teams can change people’s hospital admissions in such a way that it has
a positive influence on their total experience. This doesn’t only reply in
medical cases, but matters in every situation where patients or customers are exposed
to a certain experience.

The information that is needed to change an
experience in a positive way, relies on latent needs and wishes of patients or
customers. We want to know how they want to be treated and what they are
missing in their current experience. These latent needs are very hard to
express, especially when you are in an unfamiliar or stressed situation.

In addition, both the participants as the
interrogator/guide have only a little time to spend on this patient preference elicitation.
The patients or customers want to go back to their daily life as soon as
possible and the company or organization only has a limited budget for these

Currently, conventional user studies are
used because of their high efficiency when talking about time and budget. These
studies uncover explicit and observable knowledge from people’s current and
past experiences, but don’t expose wishes and needs of future experiences (Preece,
Rogers and Sharp 2000).

Nowadays, generative tools are emerging as
the solution for accurate preference election. These techniques uncover
people’s tacit aspirations and needs (Sleeswijk Visser, Stappers, van der Lugt and
Sanders 2005) (Sanders and Stappers 2012). Unfortunately, these studies are
known for their high effort and time-consuming aspects. In this paper, a generative
introduction to a preference elicitation study is proposed as the solution for
the patient preferences problem. 

There are three important requirements for large
scale patient preference electing. The study needs to be accurate, quick and
budget friendly.


We want participants to express their
latent needs and wishes as good and detailed as possible. Talking about these
subjects and explaining them can be very hard. The participant might not even
be aware of having them. The preference elicitation should be drawn up in such
a way that participants can express their latent needs and wishes, and that
they do it accurately. The accuracy of their responses is hard to verify. We do
can make sure that the participants are provided with questions or tools that
make it easier to express their needs and wishes accurately.


Companies or organizations often have a little
budget for carrying out patient studies. This budget covers both material costs
as personnel costs.


Not only the company or organization, but
also the participants want to go through the study as quick as possible. In the
view of the investigator, time is money. So, the quicker it is done, the less
money they have to spend. The participant just wants to get past the situation
as soon as possible. Hospital situations by example, are far from pleasant and
the patient want to put the experience behind him and go home. 

Choosing an appropriate tool for eliciting
preferences is always a balance between these three factors.


Generative tools
are emerging in cases where researchers or designers want to learn from people,
users, visitors, patients, … It’s a people centered approach where the
participant has the role of the expert and the researcher or designer is just
there to observe and to trigger thoughts and expressions. By going through a
specific process of expressions and creating an artefact, the participants are
able to express their latent feelings and thoughts, in both a visual and verbal
way. The participants are often provided with a toolkit that supports this process
of thoughts and the creation of an artefact (Sanders 2000) (Stapers and
Sleeswijk Visser 2011). The path of expression and creating an artefact are two
main factors of generative tools that cannot be found yet in conventional user
studies. In conventional user studies, the participants are missing the
triggering process of thoughts and the opportunity to visually express their
feelings. Without these, it can be very hard or even impossible to express
latent needs.

Say, Do, Make

tools are often a mix of say, do and make techniques (Figure 1).

do technique is all about observing actions of participants in their daily
life. The researcher or designer focuses on the behavior of the participant and
uses photo or video cameras, notebooks, voice recorders, to do so.

the say technique, participants express their opinions, needs, interpretations
or reasons in an interview or questionnaire. The information that is received
from this technique goes beyond the superficial layer of behavior, but is
limited to the knowledge that the participant can easily recall.

The make technique goes to
a deeper level of knowledge and releases both tacit and latent thoughts or
needs that are not that easy to express. Therefore, the participant must jump
into imagining and performs a creative act to make an artifact. Toolkits are
provided to support participants in this process. The kits are trigger sets
with all kinds of materials that provoke thinking on a deeper level (Sleeswijk
Visser, Stappers, van der Lugt and Sanders 2005) (Sanders 2009)


Figure 1: Say do
and make in order of accessibility, different methods and levels of knowledge

a study about the preferences of hospital patients regarding their hospital
room, the researcher could first observe a day in the life of the participant
by following him through his habits and routine during his  stay (DO). After this observation, he could
talk with him about things that differ from his normal life at home (SAY).
Having this done, the researcher could provide the participant with a Lego
building set and ask him to create his ideal hospital room (MAKE).

Path of Expression

Another important
aspect of generative sessions is the path of expression (Figure 2). As
explained in the previous paragraphs, generative tools are intended to evoke
the awareness and expression of latent needs and thoughts. The path of
expressions helps doing so by guiding the participant in steps through what is
meaningful from his current situation, then from his past experiences, and
afterwards using that as an input for ideation about the future. It serves as a
framework for planning the path that the participant will go through in a
generative session. The path of expression method typically works as follows (Sanders
and Stappers 2012):

The participant observes and
describes his current experiences.

The participant selects and
reflects on memories of past experiences and shares them with the group or

This makes the participant able
to express latent needs and values.

The participant uses this
latent output for exploring his aspirations for the future experience.

2: The path of expression shows how awareness can be guided in steps by
thinking first of the present (1), then of the past (2), then looking for
underlying layers (3), in order to move toward the future (4).

In the same study
about preference elicitation regarding hospital rooms, the participant might be
asked to first explain his current experiences, then reflect on and share
experiences when he still lived in his own house, and afterwards use the
discovered underlying needs and values to explore possibilities of future hospital
room experiences.



Generative tools are perfect for preference
election, but when time and budget are important factors, generative sessions
are less appropriate. On the other hand, A conventional user study consisting
of a long list of questions with check boxes is very time and budget efficient,
but misses the accuracy of patient preferences. How can we combine both? Or in
other words, how can we make generative tools quick and budget friendly?


A generative study is very intensive and
takes a lot of time. First, the research part is way more extended than a
conventional user study. The researcher must not only be present at the session
to guide and support the creativity of the participant, he also need to prepare
it very well and to process de overload of data. Both the preparations and
processing afterwards, as the session itself, are way more intensive than a
general questionnaire that can be scanned by computers. These extra hours of
work need to be payed, and when it comes to large scale research, this
difference will be enormous. Here, time results in budget, which will be discussed
in the next chapter.

Not only the researcher, but also the
participant will spend more time during a generative session than in a
conventional user study. When it comes to large scale research, this can be an
issue. The research team wants to involve many patients, users, clients, …, and
therefore, the threshold should be low.

In order to make generative sessions work
for large scaled research, its time must be reduced, both for the researchers
as the participants. But how can this be done? First, the participant should be
able to fulfill the generative study without the presence of a researcher.
Second, the preparation of the materials and processing of the data have to be


The equipment of a normal generative
session is very extended. Many triggering objects like papers filled with
pictures or words, post its, magazines, puppets, toolkits … are used during the
session. The material costs are not the only concern about this equipment, also
the storage and preparation of the materials have to be taken into account. If
we want to implement generative sessions in big preference studies, we need to
find a way to deal with the extra materials that are required. They must be as
cheap and limited as possible, without losing the triggering effect.

sum up, we want a quick and budget friendly generative session that evokes
preferences accurately in large scale studies. Therefore, the role of the
researcher should be limited, as well as the effort of the participant. The
amount of supporting materials should be reduced as well, without losing the
triggering effect and keeping in mind the path of expression and the say, do
make technique.


To respond this need for a
generative tool that fits large scale preference studies, I propose a
conventional user study with the flow of a generative session. In practice,
this implies a questionnaire that is composed, based on the general theories of
generative tools, which are a combination of say, do, make and the path of

The path of expression can
be involved by asking the questions in a specific order: going from the present
to the past, and further to the future. This specific order alone will not make
the participant able to express all his latent needs and feelings, but it does
help him to reason and argue his future wishes. Since the research team is
looking for innovation in the future, the focus will be on the last stage of
the path of expression. Although, the first two stages, past and present, are
useful to get to know the target group and to gain empathy with it.

Because of the focus on
large scale research and the absence of researchers during contact with the
participant, I suggest leaving out the ‘do’ part and make it a ‘say and make’
exercise. The ‘say’ part can easily be involved in a questionnaire. The ‘make’
part is rather hard to involve, and specially to process these overloads of
data afterwards. As a solution for this, the making of an artefact can be
involved in the first to stages of the path of expression. These stages are
mainly important for the participant to warm up and to find links between his
current and past experiences, so he can accurately describe his desired future
experience, which is of interest for the research team. The research team doesn’t
necessarily have to go through all these present and past descriptions. They
can randomly pick out a few to get some insights in the target groups
experiences and gain empathy. The data of the ‘future’ stage is essential and
can be collected and processed via a digital database.

in fact, the first two stages are an introduction to the final questionnaire
about future experiences, that helps the participant framing and reasoning, and
that provides the researchers documents and artefacts they can go through to
get inspired and to empathize.

In practice, the make part that must be
involved in the present or past experience description, can be a collage making
question. In a research about the experience of hospital patients, they can ask
the following: ‘Describe how you felt in the waiting room of the hospital,
using the supplied stickers.’ The participants will be provided with sticker
sheets filled with illustrations, pictures and words. The creation and
reflection of this artefact helps the participant to express his latent
feelings and needs, which can be used to formulate wishes and preferences in
the following questionnaire. 

To sum up, I propose the implementation of
two introductory phases about present and past experiences before going to the
final questionnaire about preferences for a future experience. The introductory
part includes the making of a collage, that helps the participant to link
specific feelings of the past and the present and express his preferences for
the future.


Without the researcher being present during
the study, it is improbable to truly understand the pattern of thoughts and
feelings of the participant. There is a lack of empathy, which has a negative
influence on the power of the research (Kouprie and Sleeswijk Visser 2009).

The inability of the participant to
verbally explain his artifact, also reduces the value and depth of the
expressed preferences afterwards. By sharing his findings out loud, the
participants get triggered to make more links between experiences, which makes
him more able to express latent needs (Sanders and Stappers 2012).

So, is what remains still a decent
generative session? The answer is no. In fact, it is not even a generative
session at all. It is still a conventional user study, but inspired on the
richness of a generative session. It is a conventional user study with the flow
of a generative session. This ‘flow’ makes it easier for the participant to
reason and argue about his preferences. 


When looking for innovation and future
opportunities in large scale preference research, conventional user studies
often don’t represent accurate and profound preferences. Generative sessions
could lead to this kind of information, but are often to intensive and
expensive to implement in large scale studies.

The solution that is proposed in this paper
is the implementation of a generative introduction to a preference survey. The
generative introduction consists of a few questions about present and past
experiences. At least one of these questions need to be answered by making an
artefact, such as a collage, using supplied stickers of pictures and words.
Going through this path of expression and making an artefact will help the
participant to link meaningful feelings and argue them.

There needs to be awareness of the possible
lack of empathy between researcher and participant, since the researcher won’t
be present during the completion of the questionnaires. Consequently, the
participant fulfils the survey on his own and don’t has the opportunity to
verbally share his findings, which will influence the outcome of his
preferences. Therefore, this technique is no generative tool but still a
conventional user study, inspired on the richness of generative sessions.

This paper describes a theoretically
construction of a new, hybrid method. Further research on the quality of this
technique and its added value needs to be done. A study where the outcome of a
conventional user study, a generative session and the new, mixed technique are
compared, is a needed next step to reflect on the proposed solution of this


Dekkers, T.  (2017) Personal

2.   Preece, J., Rogers, Y. and Sharp, H., (2000) Interaction Design:
Beyond Human-computer Interacton (Wiley: New York)

3.   Sleeswijk Visser, F., Stappers, P.J., van der
Lugt, R., Sanders, E.B.N. (2005) Contextmapping: Experiences from practice. CoDesign, 1(2), 119-149

4.   Sanders, E.B.-N. & Stappers, P.J. (2012) Convivial Toolbox.
Amsterdam: BIS Publishers

5.   Sanders, E.B.-N. (2000) Generative tools for co-designing, Collaborative design

Stappers, P.J., & Sleeswijk
Visser, F. (2011) Context & Conceptualization Reader. Faculty of Industrial
Design Engineering, TU Delft.

Sanders, E.B.-N. (2009)
Exploring Co-creation on a large scale. In: Stappers, P.J. (Ed) Designing for,
with, and from user experiences, Delft: StudioLab Press, 38-42

Kouprie & Sleeswijk Visser,
F (2009) A framework for empathy in design: stepping into and out of the user’s
life. Journal of Engineering Design
20(5) 437-448