or

f b

Available online 11 March 2015

Sustainable agricultureRoles

n ansustand

but also the design of the concept of sustainability; (2) all participants (users & designers) have a role in

particie Frenc

units (environmental safety, economic viability, and social equity,usually acknowledged as the three pillars of sustainability). It thusgenerated elementary criteria to rate the sustainable potential of

tool required largee scale of a farm.ory design processin understandingtion. In this sense,s of participation

ch in ergonomics,investigating collaborative design e or co-design e activitiesoccurring in design meetings and participatory design situations(e.g. Darses et al., 2001; Detienne, 2006; Visser, 2006, 2009). Here,design is viewed as a socio-cognitive activity, which is analyzedthrough verbal interactions between participants in meetings.

* Corresponding author. INRA, UMR 1326 LISIS, Ba^timent EGER, BP 01, F-78850Thiverval-Grignon, France. Tel.: 33 1 30 81 53 56; fax: 33 1 30 81 59 39.

E-mail addresses: ore.barcellini@cnam.fr (F. Barcellini), lorene.prost@grignon.inra.fr (L. Prost), cerf@agroparistech.fr (M. Cerf).

1

2 Collaboration is coordinated, synchronous activity that is the result of acontinued attempt to construct and maintain a shared conception of the designproblem (Roschelle and Teasley, 1995).

Contents lists available at ScienceDirect

Applied Ergonomics

journal homepage: www.els

Applied Ergonomics 50 (2015) 31e40Joint rst authors.oping sustainable agriculture is currently a strong challenge in theagricultural world since the productive model of agriculture hasbeen linked to a depletion of natural resources (energy, soil, bio-diversity) with severe degradation of the environment (water andair mainly) (Millennium Ecosystem Assessment, 2005). The toolthese scientists designed e MASC, for Multi Attribute Sustain-ability Cropping systems e was a decision tree that broke thesustainability assessment decisional problem down into simpler

farmers were not directly targeted as use of theamounts of data which were not available on thThe designers' group therefore set up a participatfor MASC and asked us afterwards to assist themand transforming this participatory design situaour approach was to characterize the actual formin design meetings.

To do so, we grounded our work in resear2Agricultural Research (INRA) who were designing software tosupport the assessment of cropping system sustainability. Devel-

working on innovative cropping systems and advisors whoaccompanied farmers in examining their practices. Individual1. Introduction

Our research is focused on ainvolving agricultural scientists of thhttp://dx.doi.org/10.1016/j.apergo.2015.02.0050003-6870/ 2015 Elsevier Ltd and The Ergonomicsco-designing the concept (in our case, sustainability); (3) some roles and proles are key to this co-design. We discuss our contributions to both the research and the practices of participatory design.These contributions deal with the production of a method and related knowledge about actual activitiesin participatory design situations. They may support the development of relevant training programsregarding participatory situations, or be reexive activities that can help those who are involved indesigning and leading in participatory situations, to make improvements.

2015 Elsevier Ltd and The Ergonomics Society. All rights reserved.

patory design processh National Institute for

cropping systems (Sadok et al., 2009). The designers' aim was tomake the concept of sustainability more tangible and easier tochallenge by those involved in the assessment of current or newcropping systems. These targeted users were mostly scientistsKeywords:Participatory designet al., 2013) to capture the levels of abstraction (conceptual, functional and operational) of partici-pants' discussions. We show that: (1) the process does not only concern the design of the artifact itself,Designers' and users' roles in participatco-designed by participants?

Flore Barcellini a, 1, Lorene Prost b, 1, *, Marianne Cera Ergonomics Team, CNAM, CRTD, 41 rue Gay-Lussac, 75005 Paris, Franceb INRA, UMR 1326 LISIS, F 77420 Marne-La-Vallee, France

a r t i c l e i n f o

Article history:Received 7 August 2014Accepted 17 February 2015

a b s t r a c t

This research deals with asoftware that assesses theparticipation of designersSociety. All rights reserved.y design: What is actually

alysis of forms of participation in a participatory design (PD) process of aainability of agricultural cropping systems. We explore the actual forms ofusers by adapting an Actual Role Analysis in Design approach (Barcellini

evier .com/locate/apergo

These interactions are seen as the traces of the participants' activ-ities in these meetings. This approach reveals how the negotiationof various perspectives involved in collaborative design processes(Bucciarelli, 1998; Detienne et al., 2005; Baker, 2009) contributes tobuilding design solutions (their generation, exploration, compari-son, and selection). Analyzing the exchanges that result from theconfrontation of perspectives is particularly relevant in participa-tory forms of design. In line with participatory ergonomics pro-posals (Haines et al., 2002; Rocha et al., 2015), ergonomists focusingon design activities have often emphasized the need to involve thepotential users of the artifact under design (whether it be softwareor work organization) in the design processes. While there are richdiscussions in the ergonomics literature about the objectives andmethods of participation in design processes (see for instance

adapting the ARAD approach. We then describe and explain howour data were collected and processed, before presenting and dis-cussing our results. We nally discuss the contribution of ourresearch to the development of methodologies for studying actualparticipation in participatory design.

2. Theoretical approach

2.1. Adapting the ARAD approach to identify forms of designparticipation

The ARAD approach proposed by Baker et al. (2009), Detienneet al. (2012b) and Barcellini et al. (2013) has been designed tocapture actual collaborative design activities, i.e. that are not pre-

n

F. Barcellini et al. / Applied Ergonomics 50 (2015) 31e4032Broberg et al., 2011), fewer studies have focused on the content ofthe interactions in participatory design meetings (e.g. Darses et al.,2001; Beguin, 2003; Engestrom and Toivainen, 2011). This parallelssome current challenges of Participatory Design research(Greenbaum and Loi, 2012). Participatory Design has been denedas a strong commitment to understanding practice, guided by therecognition that designing the technologies people use in theireveryday activities shapes, in crucial ways, how those activities mightbe done (Robertson and Simonsen, 2012). In this trend, the usersoccupy a very specic and demanding place in the design process,as co-designers, which goes beyond the way they are usuallyconsidered in design meetings (Nelson et al., 2013). But whatexactly do they contribute to design? What content is sharedamong users and designers in such processes? The epistemic con-tent of the design interactions is thus the subject of interest. Thispaper follows this line of investigation of participatory designmeetings. Our objective is to investigate the actual content ofparticipatory design meetings and to establish whether the par-ticipants discuss the operational dimensions of the tool underdesign or its conceptual dimensions. In other words, do all theparticipants (and which participants?) design the MASC tool/and/or the concept of sustainability which is implemented in this tool?We assume that this understanding is of particular interest to re-searchers, teachers and practitioners in ergonomics involved in thedevelopment of accurate participatory design processes. Thisobjective requires us to develop interactionist and developmentalmethodologies, using a ne-grained qualitative approach. One ofthem is the Actual Role Analysis in Design (ARAD) approachproposed by Barcellini et al. (2013) to study participation in OpenSource Software (OSS) design (see Section 2.1). We have adaptedthe ARAD approach to capture the roles of participants regardingthe actual content of their discussions, and to identify key-participants fostering participatory design processes by perform-ing specic combinations of roles (called proles). In the followingsections we rst set out the theoretical approach that we built by

Table 1The four types of role.

Interacting role Group-oriented role

Goal Embracing the level ofparticipation of a participant& his/her position in the relatedcommunication network ofdesign discussions

Characterizing participationto coordinate the group: coordination of activities

(e.g. allocation of tasks) regulation of interactions

Descriptors Number of contributions orturns in discussions

Position in interactions:opening a turn, closing a turdened but are effectively performed by participants, and thatemerge from actual interactions between participants. It identiesroles that correspond to distinctive and regular individual behav-iors emerging in the interaction. Four types of role are considered toembrace different facets of participation (Table 1). They are char-acterized on the basis of the structure of the interactions duringdesign meetings (interactive role), as well as according to theorientation of the interactions amongst the participants engaged indiscussions (group-oriented or task-oriented), and by quoting thedirect actions undertaken to modify the artifact (production role).

Applying this approach to the specic concern of this paperrequires us to adapt the characterization of the four roles proposedby Barcellini et al. (2013). These roles were dened to studyparticipation among participants interacting online (e.g. throughmailing-lists) e and not in a face-to-face context. We thereforehad to adapt the characterization of interactive and group-orientedroles to a face-to-face context. Moreover, the original task modelwas focused on the way design solutions were collaborativelydeveloped, whereas in this paper we wanted to ascertain whetherthe participants discussed operational or conceptual dimensions ofMASC. We therefore based our task-oriented role analysis on thethemes addressed by participants, rather than on their contributionto collaborative design activities (e.g. proposition or evaluation ofdesign solutions) as in Barcellini et al. (2013) (in bold in Table 1). Todo so, we categorized interactions according to the levels ofabstraction characterizing the participants' input (Pols, 2012;Rasmussen, 1986; Visser, 2006). The notion of level of abstrac-tion models the evolving representations of artifacts during thephases of the design process, frommore abstract representations tomore physical-concrete ones, and helps in considering how theparticipants progress throughout levels of abstraction (Visser,2006). We choose to refer to a specic abstraction hierarchyadapted to our case study. Thus, MASC, which is dedicated to theassessment of the sustainability of cropping systems, can beconsidered as a management instrument, i.e., an apparatus

Task-oriented role Production role

Embracing participation in the considered task Characterizingdirect actions onthe consideredartifact

Depends upon the design situation under scrutiny:- In Barcellini et al. (2013), the task model used wasthat of collaborative design activities (generation,clarication, evaluation of solutions)

- In our study: Themes grouped into levels ofabstraction and number of turns of each

Number ofmodicationsof the sourcecode of the softwareparticipant for this level (see Table 4)

conceived by organizations to lead the objects that they comprise andindividuals towards determined ends (Moisdon, 2006 in Labatutet al., 2009). From this perspective, three abstraction levels mustbe considered in the design process: the tool as a tangible object(operational level), its use in an organization for dened purposes(functional level), and a level related to the goals to be reached,linked to the managerial philosophy which is embedded in thedesign of the tool (philosophical or conceptual level) (Labatut et al.,2009). In our case study, the conceptual level is linked to theconcept of sustainability: the managerial philosophy is reected bythe choice to represent the contribution of cropping systems to

3.3.1. Construction of roles and prolesRoles are characterized by analyzing the content of meetings on

the basis of the approach described in Section 2. Proles have been

Project leader Super experts

cipation in useg-listsionsof knowledgee designmpionsccording to their specic

Interactive role: closing positionor starting position in interactionsGroup-oriented role: coordinationat the meta-level of all the software,and at the level of the piece underdiscussionDecision makingRegulation of discussions anddecision making

Interactive role: participationto a high degree in designmeetingsGroup oriented role: providersof support activities for otherparticipantsTask oriented role: integrationof knowledge from variousdomains, in particular design

Table 3Data on the agricultural scientists of the MASC group.

Initial of theagronomist

Status Field of expertise

T Professor, project manager Cropping systems, global agronomyD Engineer, project facilitator Junior agronomistL Research engineer Integrated agriculture, cropping

systemsR Research engineer Cropping and farming systemsF Research engineer Cropping systems and geneticsB Research engineer Organic farmingC Research engineer Biodiversity, agronomic and

environmental indicators

F. Barcellini et al. / Applied Ergonomics 50 (2015) 31e40 33sustainability by means of a decision tree, criteria, their weighting,and aggregation rules. The functional level relevant for MASC isrelatedwith the function of the tool, that is to say, the assessment ofcropping systems, which also needs to be framed and claried. Andthe operational level is the level where the operating of MASC isdiscussed.

2.2. Characterizing proles to identify key-participants fostering thedesign process

The notion of prole e as an aggregation of these four roles e isproposed to capture the variety of roles performed by participantsin different situations (for instance different in time or in problemsaddressed by participants). Thus, proles describe a participant interms of his/her behavior, compared to other participants (Bakeret al., 2009). It is particularly useful to understand the function ofkey-participants fostering a given design process, in our case aparticipatory design process. In this sense, these proles can bediscussed with forms of participation presented as essential tosoftware design performance: champions, boundary spanners,leaders and super-experts (e.g. Curtis et al., 1988; Sonnentag, 1998;Sonnenwald, 1996), in relation to roles. Table 2 presents the de-nition of these potential proles in relation to the roles of partici-pants, as studied in Open Source Software design by Barcellini et al.(2008, 2009, 2013).

3. Material and methods

Our methodology combines a monitoring of the MASC designprocess; the coding of the content of design meetings for the pur-pose of identifying the level of abstraction addressed and the rolesand proles of the participants in the discussions; and nally in-terviews with MASC designers to contextualize our results.

3.1. Data gathering on the participatory design process of MASC

software

A rst prototype of MASC (MASC 1.0) was produced in 2008(Sadok et al., 2009) by seven agricultural scientists. In 2009 it washanded over to several users to assess contrasting cropping systems

Table 2Denition of proles.

Proles Champions Boundary spanners

Interactive role: specic position in designdiscussions (beginning and/or closing) andbeing signicant contributors to productionand tasks

Interactive role: cross-partiand design-oriented mailinCentral position in interactTask-oriented role: transferfrom users and usage to thdiscussions, Support of chaPotentially super-experts aproblem addressed and theeld of expertisewithin diverse professional contexts (farmers' advisors, engineersand scientists from farmers' owned applied research institutes andfrom agronomic public research). The decision was made todevelop a second version of MASC and a junior agronomist washired to support the initial 7-researcher design team. Our empiricaldata deals with the design process of this second version, whichlasted fromMarch to November 2010 and which involved the teamof designers, called the MASC group (eight people, see Table 3)and users. This MASC 2.0 design sequence was opened and closedbymeetings with users (UM1 in March 2010, involving all the usersof the rst version of MASC, and UM2 in November 2010, withthree volunteer representative users: a scientist for INRA, anagronomist from an applied research institute and an advisor).Between these two meetings the designers worked together toprocess all the comments made by the users (DM 1, 2, 3) (Table 4).Our datawere collected during these vemeetings andwere audio-recorded and transcribed.

3.2. Data analysis: identifying the themes of the design discussionsand their levels of abstraction

The content of the design meetings was divided into 137 the-matic sequences (709 min) dened according to the differentthemes addressed by participants (Darses et al., 2001) (excludingthose concerning the overall organization of the design process).For each sequence, we noted who introduced the themes, whosuccessively took part in the discussion, the number of turns ofeach participant, the length of the sequence, and the themesaddressed in relation to the level of abstraction. The rationale usedto link the themes addressed and the level of abstraction isdescribed in Table 5.

3.3. Identifying participants' roles and proles

J Research director Modeling-statisticsand application domains

eetin

that assess the sustainability of agriculturalsystems were presented to open the debates; 3)

calereneetin

eetin

Ergconstructed by aggregating the specic roles performed by eachparticipant (each member of the MASC group and users) inrelation to the activities of other participants in meetings (for thisstep of our analysis we do not consider UM1 because the projectfacilitator was not yet involved in the project).

These comparisons between participants supposed certaintypes of metrics to characterize the role performed, i.e. whatcharacterizes a participant within the group (Table 6). Two wereused to construct participants' prole: position in the distributionof actions and tendency to talk at one or another level of abstractioncompared to others.

For production, interactive, and group-oriented roles, we considerthe intensity of participation and characterize the role of theparticipant (high, low, mean) regarding his/her position in thedistribution of actions (i.e. last quartile Q4 or rst quartile Q1). Forinstance, the interactive role of a participant is qualied as high if

Table 4Description of the meetings analyzed in our study.

Code of the meeting Type Length(min)

Modality

UM1 (March 2010) Use meeting (two-day seminarabout uses of MASC 1st version)

648 Physical m

DM1 (April 2010) Design meeting 126 ConferenceDM2 (May 2010) Design meeting 178 Video confDM3 (June 2010) Design meeting 187 Physical m

UM2 (November 2010) Use meeting (discussions aboutthe choices made in MASC

2nd version)

229 Physical m

F. Barcellini et al. / Applied34the number of turns she/he is responsible for is in the fourthquartile of the distribution of all turns (superior to Q3).

For the task-oriented role, we use the tendency of each partici-pant to talk at one or another level of abstraction compared to theother participants. For example, if one participant talks more at aconceptual level than all participants on average, he/she will becharacterized as high for this level. Following Barcellini et al. (2013),this metric was preferred to an evaluation of intensity of perfor-mance of a given level, because roles are aimed at characterizingthe specicity of some participants compared to the others. Thetendency is evaluated on the basis of relative deviation (RD), adescriptive statistic used for Contingency Tables, which highlightshow each modality of a categorical variable in a column is associ-ated with each modality of another categorical variable in a row(Bernard, 2003). By convention, there is a high positive (respec-tively negative) association between a participant and a themewhen the absolute RD value is 0.2 (respectively

Table 5Conceptual, functional and operational levels of abstraction addressed in meetings.

Level of abstraction Theme of discussion Justication Example

Conceptual level:concept ofsustainability

Inclusion of new criteria Choosing to add a criterion challenges theparticipants to examine whether thiscriterion is relevant for assessing sustainability

On the inclusion of a criterion of innovationand novelty: some farmers could value a systembecause it is innovative, it would give them aspecial social status. But I don't know if it ispart of sustainability.

Tree structure Building the hierarchical tree structureimplies that the participants show how theysee the contribution of the cropping systemsto overall sustainability

On the branch environmental sustainability:is it a good choice to cut it into two sub-branches:for society and for farmers? This would mean thatthere is an environment that provides productiveservices for farmers and an environment thatprovides services for society. []But this distinction

ay thts co

s coo nueed

F. Barcellini et al. / Applied Ergonomics 50 (2015) 31e40 35Weighting/aggregationrules

These rules make visible the wunderstand sustainability and i

Function level:assessingsustainability

Goal of the assessment Assessing how cropping systemsustainability can correspond tTo fulll its function, the tool nWhy do we observe such centrality of the conceptual level in allthe discussions? First, it is consistent with the nature of the MASC

group. As scientists, the members of this group are more likely toquestion the concepts embedded in the software. The very conceptof the tool is new enough to be questioned, and the conceptof sustainability is sufciently controversial to trigger these

adapted to themDiscriminating capacityof the tool

The assessment produced by MASConly if it allows the users to clearlybetween cropping systems

Generic nature of the tool Proposing one standard version orlinked to the goal of the assessmen

Calculation of the criteria The way the criteria are calculatedcapacity to fulll its function

Operational level:operating the tool

Tool exibility This exibility raises questions abotool to be adequately used within v

Data required to makethe tool work

The data needed are discussed in repossibilities for the users to collect

Use situations of the tool The idea of adapting the operationdiversity/specicity of the use situa

Instructions for use The participants debate on the docto ensure proper use of the tool

Table 6Synthesis of rules used to construct proles of participation.

Adapted from Barcellini et al., 2013.is not so easy.e participantsmponents

When we look at the adjustments made by the users,weighting can vary and the differences may exceed 50%.It might be linked with each user's context and notonly with their idea of sustainability.

ntribute tomerous goals.s to be

Some users may have very strong local stakes, forinstance those in a drinking water catchment area,and this will inuence what they want to obtain from

conceptual discussions. Secondly, the conceptual level is alsoconsistent with what the group had imagined about the role of theparticipants who use the software. From the interviews that weconducted with them, it appears that the members of the groupthought that discussions about the use of the tool would help themto question the concept underlying it. The intensity of the

using MASC . is informativediscriminate

J will present a sensitivity analysis which examinesthe effects of changes in the three or in the weightingin the assessment of the overall sustainability.

a series of trees ist

Some users explained that they had modied theweighting to emphasize specic objectives of theirassessment or to represent the diversity of objectivesin their area. Most of them tried two weightings of theoverall sustainability: one favoring the economicbranch and the other favoring the environmentalbranch, to nd a balance.

impacts MASC's How can we calculate the indicator about long-termfertility of the elds? To solve this question, we have todene the meaning we give to this notion and its rolein sustainability

ut the capacity of thearious use contexts

In view of the diversity of reasons that led the users tomodify the weighting of the tree, it is difcult to imaginea new absolute proposition. There is no convergence.So we should keep one standard version and, in theinstructions for use, propose some help to modify the tool.

lation to the practicalthem

We have to respect a practical consideration: the datanecessary to estimate this criterion should be easilyavailable or readily determinable.

of the tool to thetions

The tool could be used as a tool of debate, but it must beadapted, for instance, to support these debates.

uments/help needed In the practical guide that accompanies the tool, we shouldtell them that modifying the tool is not insignicant. And itwould be good to illustrate this with examples.

Fig. 1. Themes of the discussion by level of abstraction when the agronomists are alone (DM) or not (UM) and in duration (minutes).

F. Barcellini et al. / Applied Ergonomics 50 (2015) 31e4036discussions on the conceptual level, particularly within the MASC

group, may be a good sign that they question the concept of theirtool on the basis of what they learned from the users. Thirdly,discussion at conceptual level may favor a better understandingamong all the participants and facilitate the establishment ofcommon ground between participants (Clark and Brennan, 1991).Indeed, it has been shown in more classical participatory designsituations (design of a mechanical product in industrial context)that the choices of themes addressed in meetings can integrate orexclude the users (Cahour, 2002). In our case, users are actuallyinvolved in the conceptual level of abstraction, which seemsfavorable for the group and for the enrichment of the design dis-cussions, notably because this involvement intensies cross dis-cussions between the different levels of abstraction.

4.2. Distribution of roles within participatory design meetings

4.2.1. A group and discussion management dominated by twoparticipants

Fig. 2 represents the distribution of turn-taking in design (gray)and use (black) meetings, used to determine the interactive roles of

participants. Fig. 3 represents the distribution of participants in

Fig. 2. Distribution of turns betwopening and ending positions in discussions, used to reveal group-oriented roles.

Analyzing these gures reveals the specic roles of D, the juniorproject leader, and T, the initiator of the project. Fig. 2 shows thathalf of the turns taken by participants in all discussions (design anduse meetings) (50% or 513 turns out of 1025: 513/1025) are by T andD. Fig. 3 shows that D is the participant who opens or closes asequence in discussions more frequently (45% or 93 out of 207turns in opening and ending positions, in both UM and DM: 93/207).

All the other designers (L, B, F, C, R) contribute at the same levelin discussions (from 8% to 12%, that is, 81/1025 to 127/1025) (Fig. 2).Only one participant (J) appears to participate less in discussions(fewer than 1% of interactions, 10/1025). Among them only L, B andF have a group-oriented role as they sometimes appear in begin-ning/ending positions in discussions, particularly in use meetingsfor F (Fig. 3). This may reveal some specialization of those partici-pants during discussions, which can be claried by the analysespresented below.

Finally, the users are also prolic contributors as they accountfor 40% (175/433) of the interactions during the use meetings

(Fig. 2), and are quite often in beginning/ending positions in

een participants in meetings.

Fig. 3. Distribution of opening and ending position in discussions during meetings.

F. Barcellini et al. / Applied Ergonomics 50 (2015) 31e40 37discussions (40%, 38 beginning or ending positions of the total of 97positions in UM) (Fig. 3). This shows that they have an active role inthe regulation of the group and the task.

4.2.2. A distribution of participation at the three levels ofabstraction

Fig. 4 represents the distribution of turns referring to one oranother level of abstraction in discussions for each participant. It isused to dene the task-oriented role.

Fig. 4 reveals that all participants (except J) addressed themes inreferences to all three levels of abstraction. This means that theusers did also participate in interactions on the concept and notonly in interactions on the operational level. In fact, the users havea distributed style of interaction as 42% (153 users' turns out of 360:153/360) discuss the conceptual level, 34% (121/360) the opera-tional level, and 24% (86/360) the functional level. In line with re-sults discussed in the previous section, the scientists (D, T, B, L, C, R,F) contributed more or less at the same level of abstraction and inequivalent proportions, with an over-representation of interactionson the concept (72e77% of all their interactions), except for J whoparticipated less than the others and only at the operational level.Fig. 4. Participants' contributions to concep4.2.3. Identication of ve proles supporting the participatorydesign process

Fig. 5 represents an analysis of each participant's prole. It isbased on the rules described in Table 6. Black is used for a highlevel of participation in a role, gray for a mean level of partici-pation, and white for low level of participation.

This analysis of proles helps in clarifying ve proles ofparticipation that support co-design within this PD process.

The rst prole is that of D (the junior project leader): he hasstrong group-oriented and interactive roles and is the only onewhohas a production role (by coding the software). Another specicityof his prole is that he tends to contribute less to the operationallevel of abstraction of the task-oriented role than do the otherparticipants. This reveals a role-complementarity as this part of thetask-oriented role is taken on by users C and J. D also appears tohave a strongly group-oriented role through his coordination actionin discussions. In this sense, he was a facilitator of discussionssupporting the design process. His prole is similar to that of achampion (Barcellini et al., 2008; Sonnenwald, 1996), for his pres-ence seems crucial to the performance of the design process. Themain difference is that D is formally employed for this task,tual, functional and operational levels.

their prole is similar to that of boundary spanners in so far as they

ant

Ergwhereas champions in OSS are the promoters of their ideas andattribute the facilitation of the process related to that idea tothemselves (Barcellini et al., 2008; Crowston et al., 2007). This maybe one explanation for D's mean position regarding the levels ofabstraction: he is not a large contributor to the various levels ofabstraction since he may not be as invested in the promotion of thesoftware as he would have been if it had been his idea.

The second prole is that of T who appears to second D in his task.T has a similar prole (highly interactive and group-oriented roles),but he does not contribute to the code. This prole is indeed verysimilar to that of project leader or initiator in OSS, i.e. participantswho are strongly present in discussions and have a strategic posi-tion, dealing with major design orientation choices, but delegatingsome parts of the design to other participants (mostly super-expertsand champions) (Barcellini et al., 2008).

The third and fourth proles are those of a super-expert group (L, F,R, B, C, J) which may be split into two sub-groups. These experts'inclination for one or another prole is explained by their experienceor position regarding the design of MASC, on which we obtainedcomplementary information through the individual interviews:

L, F, R and B have mean but complementary proles. B talks

Fig. 5. Synthetic view of each particip

F. Barcellini et al. / Applied38more than other super-experts at the operational level as he isthe only agronomist who has actually used the software. L ismore focused (than B) on the functional level as she is veryconcerned about the way sustainability can be estimated (as isR). Both B and F transfer operational issues to the design group,but F does so in the use meeting whereas B does so in designmeetings only.

C and J have a prole marked by a strong emphasis on a speciclevel of abstraction. C tends to contribute more to the opera-tional level than do the others: he had a strong impact in dis-cussing the use eld of the tool, but mainly in design meetings. Jalso has a very specic prole as he is a large contributor to theoperational level, whereas he participates in few sequences ofdesign meetings. Actually, he is the expert in sensitivity analysisin the MASC group: he is the one who studied how the toolreacted to changes in the elementary criteria or in the aggre-gation weights. He therefore intervenes mostly at the opera-tional level, which relates to changes that users can make in thestructure of the tool.

Finally, a last prole is that of the users. They have a high group-oriented role (as they open and close sequences duringrelay operational or functional knowledge in discussions withdesigners, and are in a specic coordination position (e.g. Barcelliniet al., 2008). This boundary spanner prole shows that the MASC

has succeeded in attributing a stronger contribution to users in thedesign process than they are often shown to have (e.g. Barcelliniet al., 2008). They nevertheless remain in an in-between position,still not at the very heart of the design process.

5. Discussion and perspectives: contributions for researchand practice of participatory design

The research proposed in this article offers methodological andepistemic contributions dealing with actual activities and roles ofparticipants in participatory design meetings (Section 5.1), but wediscussions), which gives them a strong coordination position.Regarding their task oriented role, they contribute more thanothers to the functional and operational levels. It is also interestingto note that they contribute as much as the others to the conceptuallevel of discussions, which conrms that they are important in co-designing not only the artifact but also its concept. In this position,

's participation: participants' proles.onomics 50 (2015) 31e40assume that it also has some practical contributions. These weexplicitly reveal in Section 5.2, in line with the proposals of Chungand Shorrock (2011) to bridge the research-practice gap.

5.1. Contributions to research on participatory design: methods andknowledge about actual activities in participatory design situations

5.1.1. A generic method to produce knowledge on actualparticipatory design activities?

This study provides some guidelines on amethodological way ofstudying and formalizing actual activities occurring in participatorydesign. It provides a new example of use of the Actual Role Analysisin Design, already tested in architectural design (Baker et al., 2009),OSS design (Barcellini et al., 2013; Detienne et al., 2012b), and thedesign ofWikipedia (Detienne et al., 2012b; Freard et al., 2010). Thisresearch opens an alternative way of characterizing participation indesign, by focusing explicitly on the level of abstraction in whichthe participants are engaged regarding the design artifact. Webelieve that the approach has proved not only its generic interestfor analyzing actual design activities in real-world design practice,but also its exibility as it can be adapted to answer questionsspecic to different PD research contexts.

Erg5.1.2. A focus on the co-design of a concept to reinforce the politicaldimension of participatory design?

The present study examines the actual activities of participants(users and designers) in a participatory design process, with theidea of more specically investigating how the concept of design isdiscussed and stabilized and by whom. The importance of theconceptual level of discussion was demonstrated in other designsituations aswell, such as in Online Epistemic Communities. In theirwork on collaborative design of an article related to Astronomy, onthe FrenchWikipedia, Detienne et al. (2012b, in revision) reveal thatover and above the sharing of different types of knowledge, a realissue is the way one conceptualizes the encyclopedia: a resource foreveryone or a resource bringing together the most accurate, preciseand recent knowledge? This leads to conicts between the types ofknowledge to be reected by an article: astronomical knowledgecontributed by scientists, and everyday knowledge contributed byenlightened amateurs. In the MASC case we reveal that, by settingup a PD process, the agronomistsedesigners create a situationwhich allows users and designers collaboratively to design theconcept sustainability of agricultural systems at the same time asthey are creating the tool. That helps participants in meetings todiscuss whether the contribution of cropping systems to sustain-ability should be considered and how. Discussing the concept heremeans that participants in meetings (users and designers) areallowed to dene orientations for the future, that is, to enhance topolitical side of participatory design (Beck, 2002). In our case, bydesigning one decision tree rather than another, one selects a visionof agriculture and the individuals called users are granted the rightto take part in this selection.

5.1.3. Using proles to analyze participatory design activitiesWe reveal ve proles of participation supporting the perfor-

mance of participatory design discussions. They are close to thoseoutlined in the design literature (champion, boundary spanners,leader and super-experts), particularly in the OSS context. As in OSS(Barcellini et al., 2008; Crowston et al., 2007; Mockus et al., 2002),there is a core team structured around the super-experts, theinitiator of the project and a facilitator. Around this core team thereis room for users' participation regarding not only usage issues(operational level, for instance), but also conceptual issues. This isin line with the OSS literature which shows that it is a key elementof the success of OSS design (Barcellini et al., 2008; Raymond, 1999;Scacchi, 2001). In our case study, however, experts are so far agri-cultural scientists and it seems that the development of anextended core of experts should be promoted in order to enlargeco-design processes (for instance to promote constructive conictin the sense of Baker, 2009). Likewise, we do not identify clearproles of boundary spanners (Barcellini et al., 2008) articulatingconceptual and functional/operational levels. This means that noparticipant involved in all design meetings transfers knowledgeabout functional/operational issues in these meetings; only theusers may do so but they are not present in all meetings.

5.2. Contribution to the practice of participatory design andparticipatory ergonomics: supporting the development ofparticipatory situations

The contribution of our research to the practice of participatorydesign and participatory ergonomics is twofold. First, the researchcan contribute to improving the teaching of participatory ap-proaches in ergonomics training programs. Second, we discuss theuse of models of actual activities as tools to enhance the reexiveactivities and learning of stakeholders involved in designingparticipatory situations and consequently to support the develop-

F. Barcellini et al. / Appliedment and quality of the situations proposed.5.2.1. Development of participatory practices in ergonomicstraining programs

Teaching the practice of ergonomics requires the construction ofpragmatic skills, in order to help trainees carry out actual practicesregarding transformative actions (Gadbois and Leplat, 2004; CREE,2007). Action theory (Schon, 1983) stresses that practitionerscannot just apply pre-dened plans of action while facing a givensituation. This is particularly relevant when building participatorydesign situations to support work transformation (Barcellini et al.,2014; Barcellini and Van Belleghem, 2014). Training programsshould therefore support the construction of competences to selectand to adapt participatory situations to the diversity of contexts(Gadbois and Leplat, 2004). This requires an accurate understandingof what occurs in real-world participatory situations. In this sense,this research can help in enhancing the understanding of the par-ticipants' contribution to participatory activities and the roles ofparticipants that are necessary to build an efcient participatoryprocess.

5.2.2. Development of reexive activities to enhance the evolutionof participatory situations

A second contribution of this type of research is its potential rolein supporting the reexive activities of participants involved indesigning and leading participatory design situations. Organizingreexive activities (e.g. Mollo and Falzon, 2004; Mollo andNascimento, 2014) is viewed here as a tool to enhance thelearning of stakeholders involved in participatory situations andthen the organizational choicesmade to lead these situations. In ourresearch, this was based on discussing a model of actual activities tofeed reexive work and thus to answer the initial question of theresearchers-designers of MASC, to help them in enhancing theirparticipatory situations. This approach is particularly important inthe agronomic eld, in which participatory design approaches aregrowing rapidly and stakeholders are seeking this kind of support.We began to initiate a discussion with the MASC group about theresults of this study, in particular regarding the importance ofexpanding the core of participants and of boundary spanning ac-tivities. The discussion about levels of abstraction made themrealize that their position regarding the users required clarication.That opened debatese currently still activee on theway tomove tothe following steps of the process. Such a reexive approach, to becontinued, may improve the quality of collaboration (Detienneet al., 2012a) and eventually the quality of the design process andof the designed artifact. This places this current research in aresearch stream that tends to lend more interest to constructivelearning processes in design (e.g. Beguin, 2003; Engestrom andToivainen, 2011; Barcellini et al., 2014).

6. Conclusion

This paper addresses an interesting challenge: how can westudy participation and at what level of abstraction of the artifactdo participants co-design? We investigated it by following a real-world case study of a participatory design process over a six-month period. To analyze this case study, we adapted the ActualRole Analysis in Design approach of Barcellini et al. (2013). Tohighlight the discussions occurring at the level of the concept underdesign, we modied the way the task-oriented role is analyzed, tofocus on the themes and levels of abstraction that are addressed byparticipants. Although an in-depth understanding of the peopleand the subject is necessary for interpreting roles, this way ofanalyzing participation allows us to advance in our understandingof a participatory design process by showing that, in our case-study,the participation is linked with wide discussion of the concept of

onomics 50 (2015) 31e40 39the design. The participation of a broad range of people clearly

shows the co-design of the concept of agricultural systems' Darses, F., Detienne, F., Falzon, P., Visser, W., September 2001. A Method for Ana-lysing Collective Design Processes. Research report n 4258. INRIA.

Detienne, F., Martin, G., Lavigne, E., 2005. Viewpoints in co-design: a eld study in

F. Barcellini et al. / Applied Ergonomics 50 (2015) 31e4040contribution to sustainability. This reinforces the need for thoseimplementing the PD process to set up spaces inwhich the conceptand the different levels of abstraction can be discussed.

Acknowledgments

We are grateful for the nancial support of the French NationalResearch Agency (Project POPSY, grant number ANR-08-12-STRA).

We thank Wael Benkerrour who collected the data and carriedout a rst analysis of them, and Liz Libbrecht for language editingthe English version of this paper.

We are deeply grateful to all the participants of the MASC

group for their interest and availability, and extend special thanksto D who devoted a great deal of time to our work.

Finally, grateful thanks are due to the two anonymous reviewersand to the Journal Editor. Through their in-depth comments, theyprovided us with very valuable advice to improve the rst versionof this article.

References

Baker, M.J., 2009. Argumentative interactions and the social construction ofknowledge. In: Mirza, N.M., Perret-Clermont, A.-N. (Eds.), Argumentation andEducation: Theoretical Foundations and Practices. Springer Verlag, Berlin,pp. 127e144.

Baker, M., Detienne, F., Lund, K., Sejourne, A., 2009. Etude des prols interactifs dansune situation de conception collective en architecture. In: Detienne, F.,Traverso, V. (Eds.), Methodologies d'analyse de situations cooperatives deconception: le corpus MOSAIC. PUN, Nancy.

Barcellini, F., Detienne, F., Burkhardt, J.M., 2008. Users and developers mediation inan open source software community: boundary spanning through crossparticipation in online discussions. Int. J. Hum. Comput. Stud. 66 (7), 558e570.

Barcellini, F., Detienne, F., Burkhardt, J.M., 2009. Participation in online interactionsspaces: design-use mediation in an open source software community. Int. J. Ind.Ergon. 39, 533e540.

Barcellini, F., Detienne, F., Burkhardt, J.-M., 2013. A situated approach of roles andparticipation in open source software communities. Hum. Comput. Interact. 10.1080/07370024.2013.812409.

Barcellini, F., Van Belleghem, L., Daniellou, F., 2014. Design projects as opportunitiesfor the development of activities. In: Falzon, P. (Ed.), Constructive Ergonomics.Taylor and Francis, NY, USA, pp. 150e163.

Barcellini, F., Van Belleghem, L., 2014. Organizational simulation: issues for ergo-nomics and for teaching of ergonomics' action. In: Broberg, O., Fallentin, N.,Hasle, P., Jensen, P.L., Kabel, A., Larsen, M.E., T.Weller (Eds.), Proceedings of theHuman Factors in Organizational Design and Management e XI and NordicErgonomics Society Annual Conference e 46.

Beck, E.E., 2002. P for political: participation is not enough. Scand. J. Inf. Syst. 14 (1)article 1.

Beguin, P., 2003. Design as a mutual learning process between users and designers.Interact. Comput. 15 (5), 709e730.

Bernard, J.-M., 2003. Analysis of Local or Asymmetric Dependencies in ContingencyTables Using the Imprecise Dirichlet Model. Paper presented at the 3rd Inter-national Symposium on Imprecise Probabilities and Their ApplicationsISIPTA'03, Lugano, Switzerland, July 14e17.

Broberg, O., Andersen, V., Seim, R., 2011. Participatory ergonomics in design pro-cesses: the role of boundary objects. Appl. Ergon. 42 (3), 464e472.

Bucciarelli, L.L., 1998. Engineering design process. In: Dubinskas, F. (Ed.), MakingTime: Culture, Time and Organization in High Technology. Temple UniversityPress, Philadelphia, PA, pp. 92e122.

Cahour, B., 2002. Decalages socio-cognitifs en reunions de conception participative.Le Trav. Hum. 65 (4), 315e337.

Center for Registration of European Ergonomics, 2007. Harmonising EuropeanTraining Programs for the Ergonomics Profession (HETPEP).

Chung, A.Z.Q., Shorrock, S.T., 2011. The research-practice relationship in ergonomicsand human factors e surveying and bridging the gap. Ergonomics 54 (5),413e429. http://dx.doi.org/10.1080/00140139.2011.568636.

Clark, H.H., Brennan, S.E., 1991. Grounding in communications. In: Resnick, R.L.,Levine, J.-M., Teasley, S.D. (Eds.), Perspective on Socially Shared Cognition. APA,Washington, DC, USA, pp. 127e149.

Crowston, K., Qing, L., Kangning, W., Eseryel, U.Y., Howison, J., 2007. Self-organi-zation of team for free/libre open source software development. Inf. Softw.Technol. 49, 564e575.

Curtis, B., Krasner, H., Iscoe, N., 1988. A eld study of the software design process forlarge systems. Commun. ACM 31, 1268e1287.concurrent engineering. Des. Stud. 26 (3), 215e241.Detienne, F., 2006. Collaborative design: managing task interdependencies and

multiple perspectives. Interact. Comput. 18, 1e20.Detienne, F., Baker, M., Burkhardt, J.M., 2012a. Quality of collaboration in design

meetings: methodological reections. Co-Design 8 (4), 247e261.Detienne, F., Barcellini, F., Baker, M., Burkhardt, J.M., Freard, D., 2012b. Online

epistemic communities: theoretical and methodological directions for under-standing knowledge co-elaboration in new digital spaces. Work 41 (Suppl. 1),3511e3518.

Detienne, F., Baker, M., Freard, D., Barcellini, F., Denis, A., Quignard, M., 2015. Thedescent of Pluto: interactive dynamics, specialisation and reciprocity of roles ina Wikipedia debate. Int. J. Hum. Comput. Stud. (in revision).

Engestrom, Y., Toivainen, H., 2011. Co-congurational design and learning in-strumentalities: an activity-theoretical perspective. In: Ludvigsen, S., Lund, A.,Rasmussen, I., Saljo, R. (Eds.), Learning across Sites: New Tools, Infrastructuresand Practices. Routledge, London, pp. 33e52.

Freard, D., Denis, M., Detienne, F., Baker, M., Quignard, M., Barcellini, F., 2010. Therole of argumentation in online epistemic communities: the anatomy of aconict in Wikipedia. In: Neerincx, M., Brinkman, W.-P. (Eds.), Proceedings ofthe European Conference on Cognitive Ergonomics ECCE'10. Delft, TheNetherlands, 25e27 August.

Gadbois, C., Leplat, J., 2004. Knowledge and transformation actions. Activites 1 (1),6e22.

Greenbaum, J., Loi, D., 2012. Participation, the camel and the elephant of design: anintroduction. CoDes. Int. J. CoCreat. Des. Arts 8 (2e3), 81e85.

Haines, H., Wilson, J.R., Vink, P., Koningsveld, E., 2002. Validating a framework forparticipatory ergonomics (the PEF). Ergonomics 45 (4), 309e327. http://dx.doi.org/10.1080/00140130210123516.

Labatut, J., Aggeriz, F., Astruc, J.M., Bibe, B., Girard, N., 2009. The active role of in-struments in articulating knowing and knowledge: the case of animal quali-cation practices in breeding organisations. Learn. Organ. 16 (5), 371e385.

Millennium Ecosystem Assessment, 2005. Ecosystems and Human Well-being:General Synthesis. Island Press, Washington. Online: http://www.millenniumassessment.org/en/index.aspx.

Mockus, A., Fielding, R.T., Herbsleb, J., 2002. Two case studies of open source soft-ware development: Apache and Mozilla. ACM Trans. Softw. Eng. Methodol. 11(3), 309e346.

Moisdon, J.C., 2006. Sur la largeur des mailles du let. Savoirs incomplets et gou-vernement des organisations. In: Hatchuel, A., Pezet, E., Starkley, K., Lenay, O.(Eds.), Gouvernement, organisation et gestion: l'heritage de Michel Foucault.Presses de l'universite de Laval, Laval.

Mollo, V., Falzon, P., 2004. Auto-and allo-confrontation as tools for reective ac-tivities. Appl. Ergon. 35 (6), 531e540.

Mollo, V., Nascimento, A., 2014. Reective practices and the development of in-dividuals, collectives, and organizations. In: Falzon, P. (Ed.), Constructive Er-gonomics. Taylor and Francis, NY, USA, pp. 164e175.

Nelson, J., Buisine, S., Aoussat, A., 2013. Anticipating the use of future things: to-wards a framework for prospective use analysis in innovation design projects.Appl. Ergon. 44, 948e956.

Pols, A.J.K., 2012. Characterizing affordance: the descriptions-of-affordances-models. Des. Stud. 33 (2), 113e125.

Rasmussen, J., 1986. Information Processing and HumaneMachine Interaction.North Holland, New York.

Raymond, E.S., 1999. The cathedral and the bazaar. http://www.tuxedo.org/esr/writings/cathedral-bazaar/ (20.06.05.).

Robertson, T., Simonsen, J., 2012. Challenges and opportunities in contemporaryparticipatory design. Des. Issues 28 (3), 3e9.

Rocha, R., Mollo, V., Daniellou, F., 2015. Work debate spaces: a tool for developing aparticipatory safety management. Appl. Ergon. 46, 107e114.

Roschelle, J., Teasley, S.D., 1995. The construction of shared knowledge in collabo-rative problem solving. Comput. Support. Collab. Learn. 128, 69e97.

Sadok, W., Angevin, F., Bergez, J.E., Bockstaller, C., Colomb, B., Guichard, L., Reau, R.,Messean, A., Dore, T., 2009. MASC: a qualitative multi attribute decision modelfor ex ante assessment of the sustainability of cropping systems. Agron. Sustain.Dev. 29, 447e461.

Scacchi, W., 2001. Understanding the requirements for developing open sourcesoftware systems. IEEE Proc. Softw. 149 (1), 24e39.

Schon, D., 1983. The Reective Practitioner: How Professionals Think in Action.Basic Books, New York.

Sonnentag, S., 1998. Expertise in professional software design: a process study.J. Appl. Psychol. 83 (5), 703e715.

Sonnenwald, D.H., 1996. Communication role that support collaboration during thedesign process. Des. Stud. 17, 277e301.

Visser, W., 2006. Designing as construction of representations: a dynamicviewpoints in cognitive design research. Hum. Comput. Interact. 21 (1),103e152.

Visser, W., 2009. Design: one, but in different forms. Des. Stud. 30, 187e223.

Designers' and users' roles in participatory design: What is actually co-designed by participants?1. Introduction2. Theoretical approach2.1. Adapting the ARAD approach to identify forms of design participation2.2. Characterizing profiles to identify key-participants fostering the design process

3. Material and methods3.1. Data gathering on the participatory design process of MASC software3.2. Data analysis: identifying the themes of the design discussions and their levels of abstraction3.3. Identifying participants' roles and profiles3.3.1. Construction of roles and profiles3.3.2. Considered participants

3.4. Interviews with the MASC group

4. Results4.1. A centrality of discussions at the conceptual level4.2. Distribution of roles within participatory design meetings4.2.1. A group and discussion management dominated by two participants4.2.2. A distribution of participation at the three levels of abstraction4.2.3. Identification of five profiles supporting the participatory design process

5. Discussion and perspectives: contributions for research and practice of participatory design5.1. Contributions to research on participatory design: methods and knowledge about actual activities in participatory design si ...5.1.1. A generic method to produce knowledge on actual participatory design activities?5.1.2. A focus on the co-design of a concept to reinforce the political dimension of participatory design?5.1.3. Using profiles to analyze participatory design activities

5.2. Contribution to the practice of participatory design and participatory ergonomics: supporting the development of participat ...5.2.1. Development of participatory practices in ergonomics training programs5.2.2. Development of reflexive activities to enhance the evolution of participatory situations

6. ConclusionAcknowledgmentsReferences