A novel research method for workshops and co-production of knowledge: using a secret Facebook group

Abstract

Background: Co-production is reliant on good communication and consensus between participants but attending in-person meetings and workshops is hard for time-constrained groups such as new mums, who may be geographically dispersed without reliable transport. Discussions with a lay advisory group resulted in the decision to hold a workshop over a secret Facebook group. The aim of this study was to test the feasibility a secret Facebook group for co-production activities. In the example presented here, the population was women with previous gestational diabetes; the topic was physical inactivity; and the purpose was to develop an acceptable intervention to increase physical activity.
Methods: The researchers created a secret Facebook group with content similar to an in-person workshop that sequentially progressed to develop a programme theory for an intervention. The researcher posted 1-2 times per day for 14 days and members of the group were invited to comment and discuss the content. Feasibility and acceptability of the group analysed using Facebook analytics and a post-workshop survey.
Results: Twenty-one participants took part. In total, 521 comments were provided in response to 18 posts of varying types (average = 28.9 comments per post). The total word count of participant comments was 21,142 words. The workshop was viewed positively, with 20 of 21 participants saying they liked the workshop ‘somewhat’ or ‘a great deal’, and felt the group was a safe and open environment to share opinions. When asked if they would take part in something like this again, 15 of 21 said “Yes”. Participants mentioned the format was convenient; it allowed them to reflect on their own experiences and they liked helping research progress. Those who say “yes” said it was difficult finding time and depended on what else was going on.
Conclusion: Using a secret Facebook group as a method of co-production or as a workshop in the research process is a feasible and acceptable method. Social media holds significant potential for co-production and involvement in research for populations who are geographically dispersed or time-constrained; with an uncommon condition; or in other circumstances where in-person meetings are either not appropriate or not possible.

Full text

1
Preprint:Pleasenotethatthisarticlehasnotcompletedpeerreview.
Anovelresearchmethodforworkshopsandco-
productionofknowledge:usingasecretFacebook
group
CURRENTSTATUS:UND ERREVI EW
AudreyBuelo
TheUniversityofEdinburgh
Audrey.Buelo@ed.ac.ukCorrespondingAuthor
ORCiD:https://orcid.org/0000-0002-1946-6711
AlisonKirk
UniversityofStrathclyde
RuthJepson
TheUniversityofEdinburgh
DOI:
10.21203/rs.2.24083/v1
SUBJECTAREAS
SocialPolicy SocialWork
KEYWORDS
Facebook,co-production,intervention,physicalactivity,gestationaldiabetes

2
Abstract
Background:Co-productionisreliantongoodcommunicationandconsensusbetweenparticipants
butattendingin-personmeetingsandworkshopsishardfortime-constrainedgroupssuchasnew
mums,whomaybegeographicallydispersedwithoutreliabletransport.Discussionswithalay
advisorygroupresultedinthedecisiontoholdaworkshopoverasecretFacebookgroup.Theaimof
thisstudywastotestthefeasibilityasecretFacebookgroupforco-productionactivities.Inthe
examplepresentedhere,thepopulationwaswomenwithpreviousgestationaldiabetes;thetopicwas
physicalinactivity;andthepurposewastodevelopanacceptableinterventiontoincreasephysical
activity.
Methods:TheresearcherscreatedasecretFacebookgroupwithcontentsimilartoanin-person
workshopthatsequentiallyprogressedtodevelopaprogrammetheoryforanintervention.The
researcherposted1-2timesperdayfor14daysandmembersofthegroupwereinvitedtocomment
anddiscussthecontent.FeasibilityandacceptabilityofthegroupanalysedusingFacebookanalytics
andapost-workshopsurvey.
Results:Twenty-oneparticipantstookpart.Intotal,521commentswereprovidedinresponseto18
postsofvaryingtypes(average=28.9commentsperpost).Thetotalwordcountofparticipant
commentswas21,142words.Theworkshopwasviewedpositively,with20of21participantssaying
theylikedtheworkshop‘somewhat’or‘agreatdeal’,andfeltthegroupwasasafeandopen
environmenttoshareopinions.Whenaskediftheywouldtakepartinsomethinglikethisagain,15of
21said“Yes”.Participantsmentionedtheformatwasconvenient;itallowedthemtoreflectontheir
ownexperiencesandtheylikedhelpingresearchprogress.Thosewhosay“yes”saiditwasdifficult
findingtimeanddependedonwhatelsewasgoingon.
Conclusion:UsingasecretFacebookgroupasamethodofco-productionorasaworkshopinthe
researchprocessisafeasibleandacceptablemethod.Socialmediaholdssignificantpotentialforco-
productionandinvolvementinresearchforpopulationswhoaregeographicallydispersedortime-
constrained;withanuncommoncondition;orinothercircumstanceswherein-personmeetingsare
eithernotappropriateornotpossible.

3
Introduction
Co-productionofcomplexpublichealthinterventionsisseenasanincreasingnecessityin
interventiondevelopmentforimprovingacceptabilityandsustainabilityofinterventions(1).It
involveskeystakeholdersworkingalongsideresearcherstoco-developaninterventionthatis
feasible,acceptableandsustainablebytakingaccountofcontextualfactors(e.g.need,culture,
geography,preferences,resources).Co-productionininterventiondevelopmenttypicallyinvolvesin-
personworkshopsorconsultations,andfacetofaceinterviewsorfocusgroupsinwhichintervention
materialsaredevelopedandtestedwiththosewhowillreceive,deliver,orresourcetheintervention,
andfeedbackprovidedonvariousaspects.
In-personco-productionisanefficient,convenientmethodofworkingtogether,whichaids
relationshipbuilding,butisnotsuitableforalldemographics.Inparticular,creatingaworkshopor
focusgroupatatimeandlocationthatsuitseveryoneischallengingwhenworkingwithpopulations
thathaveanuncommoncondition,aregeographicallydispersed,orhavelimitedopportunitiesto
gather.However,theinternethasdramaticallychangedmodernmodesofcommunicationand
connectedthosewhomayotherwiseneverhaveanopportunitytointeract.
Facebookisanon-linesocialnetworkingsitedevelopedin2004.Originallyintendedforuniversity
students,Facebookhasnowexplodedinpopularityaroundtheworld–withnearly1.59billiondaily
users(2).Itallowsuserstogetintouchwithpeoplewithsimilarinterests,backgrounds,and
experiencesinstantaneously.ThecapabilitiesofFacebooknowextendfarpastkeepingintouchwith
friendsandfamily.Peopleuseitasamainsourceofnews,formingconnectionsaroundsimilar
hobbies,interestsandillnesses,andforself-promotion(3,4).
GroupsareafeatureofFacebookthatallowslike-mindedindividualstogethertoconnectandshare
experiences.Facebookgroupshavevaryinglevelsofprivacy:anopengroupisviewabletoanyoneon
Facebookandnotrestrictedtomembersofthegroup.Aclosedgrouphascontentonlyviewableby
thegroupmembers,whoareabletojointhroughanapprovalprocess.Anyonecanseetheexistence
ofaclosedFacebookgroup.However,asecretgroupisnotviewablebyanyoneexceptthoseinthe
group–whoareabletojoinonlybydirectinvitation(5).Itisprivateandhiddeninsearches;those

4
whoarenotmembersofasecretFacebookgroupcannotseetheexistenceorthecontentofthe
secretFacebookgroup(6).Thistypeofprivacyholdsparticularbenefitforresearch,asconfidentiality
andanonymityareethicallyimportant.TheBritishPsychologicalSocietypublishedethicalguidelines
forinternet-mediatedresearchin2017encompassing4mainprinciples:1)respectautonomy,privacy
anddignity;2)maintainscientificintegrity;3)socialresponsibilityand4)increasethebenefitswhilst
reducingtheharms(7).
Previousstudieshaveconductedfocus-groupstyleresearchinprivateorsecretFacebookgroups(8),
withthepurposeofconductinganeedsanalysisorcollectingdatainamorefocus-groupstyleto
understandaproblem.ResearchgroupsattheUniversityofIndianaandtheUniversityofEdinburgh
haveparticularexperiencewithAsynchronousRemoteCommunities,orfocusgroupsoccurringonline
inwhichparticipantsarenotonlineatthesametime,andhavepublishedwidelyonthetopic–
primarilyshortreportswithrecommendationsforfutureresearchandlessonslearned(9,10).
However,noneofthestudieswehavefoundtodateuseasynchronousremotecommunities,
particularlyinasecretFacebookgroup,toco-designaninterventionforfuturetesting.
Thispaperdescribesamethodofco-productionandinterventiondevelopmentspecificallydesigned
toovercomebarriersofin-personresearchmethods:online,secretFacebookgroups.Inthispaper,we
willdescribethedevelopmentanduseofthesecretFacebookgroupinadispersedgroupof
participantswitharelativelyrarehealthconditionanddiscussthebenefitsandlimitationsfor
research.ThissecretFacebookgroupstudywaspartofalargerinterventiondevelopmentresearch
study,usingtheSixStepsinQualityInterventionDevelopment(6SQuID)framework.The6SQuID
frameworkisbasedonsixsteps:1.Identifyingapublichealthproblemanditscauses;2.Clarifythe
modifiableandnon-modifiablecausalfactors;3.Identifyingthetheoryofchange(whatcausal
pathwaystointerruptandhow);4.Identifythetheoryofaction(howtodelivertheintervention);5.
Testandrefineonasmallscale;and6.Collectsufficientevidenceofeffectivenesstojustifymore
rigorousevaluationandimplementation.
Thepopulationgroupforourstudywaswomenwithahistoryofgestationaldiabetes,adiseasethat
affects4.4%ofpregnanciesworldwideandsignificantlyincreasesriskofType2diabeteslaterinlife

5
(11,12).Physicalactivitywasthetargetbehaviourofthisinterventiondevelopmentstudyasitisa
beneficialtoolforType2diabetespreventionforwomenwithpreviousgestationaldiabetesaswell.A
US-basedprospectivecohortstudyof4,554womenwithahistoryofgestationaldiabetesfoundthat
forevery100minutesincreaseofmoderate-vigorousphysicalactivityperformedperweek,therewas
a9%reducedrelativeriskofT2DMonset[95%CI,0.88–0.94],evenafteradjustingfordietandBMI
(13).
Alayadvisorygroupcomposedofthreewomenwithpreviousgestationaldiabeteswasconsulted
abouttheidealmethodofco-production;theyfavouredasecretFacebookgroupforseveralreasons
includingconvenience,highexistinguse,andeaseofuseofthewebsiteandphoneapplication.
Previousresearchhasshownthat81%ofmothersuseFacebook,and56%ofthesemotherscheckthe
platformseveraltimesaday(14).Additionally,onlinefocusgroupsandin-personfocusgroupshave
beenshowntobecomparableintermsofqualityofcontentcollected(15).
AimsAndObjectives
TheaimofthestudywastotestfeasibilityandacceptabilityofusingasecretFacebookgroupforco-
production.Thispaperdescribestheapproachesusedandtheirbenefitsandlimitations.
Methods
Participantsandrecruitment
Themajorityofparticipants(16/21)hadpreviouslyhadface-to-faceortelephoneinterviewswiththe
researchersinJunetoSeptember2018.Newparticipants(5/21)wererecruitedfromaFacebook
groupthatisspecificallyforwomenwithcurrentorpreviousgestationaldiabetes.
AllparticipantswerelivingintheUKduringthestudybutweregeographicallydispersed.
Contentdevelopment
TheFacebookgroupcontentwasdevelopedtoaddresssteps2through4ofthe6SQuIDframework:
toclarifymodifiableandcausalfactorsoftheproblem;developatheoryofchangeandbeginto
developatheoryofactionforaphysicalactivityintervention(1).Atheoryofchangeistheprocessby
whichchangecomesaboutforindividuals,groupsandcommunities–itasks:howarewegoingto
changethisbehaviour?Tocreateatheoryofchange,wefirstperformedasituationanalysis
(commencinginthepreviousin-personinterviewportion)anddevelopedafishbonediagramwith

6
participantstoexplainkeybarriersandfacilitatorstophysicalactivityintheirdailylives.Thenwe
aimedtounderstandmodifiableandimportantfactorstounderstandwhichfactorscanbechanged
withthegreatestscopeforimprovementintheirphysicalactivity.Finally,theresearchersanalysed
thisinformationtobegintodevelopatheoryofhowtochangetheirbehaviour:thiswaspresented
backtoparticipantsforviewsofacceptabilityandfeasibility.
Runningtheworkshop
Theworkshoptookplaceovera15-dayperiodinMaytoJune2019,thelengthofwhichwasbasedon
previousresearch(16).Toactionthe6SQuIDsteps,theleadresearchercreatedpostsforthe
Facebookpagethatsequentiallyandcumulativelyledtothestepsabovebeingcompleted.Boththe
layadvisorygroupandco-authorsreviewedthedraftsoftheposts.Theleadresearcheraimedtouse
severaldifferentformatsofFacebookpostsincludingtext,photoseries,pollsandembeddedvideos
forvarietytokeepparticipantsengaged.Emojiswereusedinasimilarmannertoparticipants.Two
layadvisorygroupmembersalsowereinvolvedasparticipantsandco-facilitatorstoencourage
discussionifpostsdidnothaveanyresponses(thoughinhindsightthiswasnotnecessary).The
researcherpostedonetotwotimesperday,withtwodaysinthemiddleoftheworkshopthatwere
‘catch-up’days(Day8andDay10).Messagesandcontentweretypicallypostedbetween15:30and
17:30GMT,assuggestedbythelayadvisorgroup,asmotherswouldbereturningfromworkandmay
haveashortbreakbeforetheireveningmeal.
Ethicalconsiderations
Confidentiality,anonymityandidentityverificationwerespecificethicalissuesconsideredin
developingtheonlineworkshop.Thenatureofsocialmediaexistingonlineholdsinherent
confidentialityrisks.Facebookultimatelypossessestheinformationonthewebsite,andparticipants
wereinformedofthisintheconsentprocess.AsecretFacebookgroupisclearlyaprivatespace–
participantsshouldfeelabletowritefreelywithoutconcernthatwhattheyaresayingisviewableto
thosewhoarenotinthegroup.Incontrasttoconfidentialityinvolvingmanagingprivateinformation,
anonymityinvolvesobscuringidentifiableinformationforparticipantsandcanbeusedtomaintain
confidentiality(17).Thepartofqualitativeanalysisthatinvolvesthehighestrisktoanonymityisin

7
disseminatingtheresults.Forresearchusingopenandonlineinternetforums,theriskstoanonymity
arehigher,asparticipantquotescanbetracedbacktotheoriginalsourceandcanpotentialidentify
theparticipant.However,thisisnotariskinasecretFacebookgroup:Asecretgroupisnot
searchableonFacebookoranysearchengine,anddetailsgiveninthegroupareonlyviewableby
groupmembers.Assuch,anonymityispreservedbarringanydatabreachesthroughparticipants
sharingorFacebooksecuritylapses.
Identityverificationholdstwoprimaryconsiderationsinsocialmediaresearch:1.Areparticipants
whotheysaytheyare?2.DoesthepersonthataparticipantportraysinaFacebookgroupaccurately
representtheirrealself?As“Facebookprofilepageamountstoablankcanvasonwhicheachuser
hasfreereigntoconstructapublicorsemi-publicimageofhim-orherself”,itispossiblethatthe
participantstheresearcherhasnotmetbeforemaynotfitintotheeligibilitycriteria(p.213)(18).
However,thisresearchisnotconcernedwithwhotheparticipantactuallyis,butratherwitheach
participant’sopinionsandperceptionsoninterventionideas.Itispossiblethatpeoplemaysaythey
approveordisproveofsomethingthattheydonotactuallyapproveordisproveof(inreallife)–thus
harmingthedatacollected–butthisisariskinanyresearchwithpeople(5).
InformedconsentwastakenonlineinaQualtricssurvey,asstudieshaveshownthatanonline
consentformprovidesequalcomprehensioncomparedtoawrittenconsentform(19).Participants
wereemailedalinktoasurveythatlaidoutthesameinformationasawrittenparticipantinformation
sheetandconsentform,butclicked‘agree’aftereachconsentstatementtoindicateconsent,and
thenprovidedidentifyingdetailstoverifytheiridentityincludingyearofbirth,yeardiagnosedwith
gestationaldiabetes,name,andFacebookprofilenameandlink.Theonlinestudywasapprovedby
theUniversityofEdinburghHealthinSocialScienceethicscommittee.
Evaluationoftheonlineworkshop
Anevaluationplanwasdevelopedtoexplorethefeasibilityandacceptabilityofthismethodto
generateatheoryofchangeandtheoryofactionfortheintervention.
Thekeycomponentsofafeasibilitystudyincludeacceptability,practicality,demand,implementation,
adaptation,integration,expansionandlimitedefficacy(20).Acceptabilityandpracticalityaremost

8
relevantforthisstudy,andhowtheywereassessedaredescribedindetailinTable1.Demandcanbe
assessedusingtheFacebookanalyticsbecauseitshowsactualuse.Implementationwillbebasedon
analyticsandtheresearcher’sownexperiencesofexecutingtheplan.Practicality,integrationand
limitedefficacywereassessedbasedontheoverallresultsoftheworkshop,whereasadaptationand
expansionshouldbeexploredinfutureresearch.
Aspectsanddesign Aimofevaluation Details
Acceptability
Survey–quantitativequalitative
Arerecipientsanddeliverers
satisfiedwiththemethod?Doesit
feelappropriate?
Surveyquestionsinvolve
assessingparticipantenjoyment,
iftheyfeltitwasasafe
environmenttosharethoughts,
thetimingandquantityofposts,
amongotherquestions.
Practicality
Survey–quantitativequalitative
Towhatextentcantheworkshop
becarriedoutwithintended
participantsusingexisting
resources(akawhattheyalready
have?)?Aretheyabletocarryout
interventionactivities?
Implementation
Memo-ingandoverallsuccessof
workshop
1.Canitbesuccessfullydelivered
toparticipantsinsomedefined
butnotfullycontrolledcontext?
2.Whatkindofresourcesare
needed?
3.Whatfactorsaffect
implementation?
4.Whatisspeed&qualityof
implementation?
Researcher’sownexperiences
(dailymemo-ingandreflectingon
process)
Demand
Facebookanalytics
1.Howlikelyisthismethodtobe
used?
2.Whatcontentgeneratedthe
mostandleastinteraction?
Actualuseoftheworkshopby
participants.SocioGraphto
measure:mostcommentedposts;
mostreactedposts;average
wordspercomment;ratioof‘seen
by’per‘#comments’
Table1.EvaluationplanofthesecretFacebookgroupworkshop.
Results
Twenty-oneparticipantswererecruitedintothestudyfrom16Mayto28May2019.Twowomenwere
recruitedfromthelayadvisorygroup,16wererecruitedfromthepreviousinterviewstageofthe
research,and5wererecruitedfromaclosed,gestationaldiabetes-specificFacebookpage.The
averageageofparticipantswas35.8years(agerange25–47)andtheaveragetimesincelast
diagnosisofgestationaldiabeteswas2.4years(range < 1to10years).Participantswerebased
throughoutScotland.Attheendoftheworkshop,oneparticipantinformedusthatshewassixweeks
pregnant(anexclusioncriteriaofthestudy),buthercommentsandresponseswerestillincludedin
theFacebookpageasshehadnotfoundoutshewaspregnantuntilthatpoint.
Responses
Table2providesasummaryofthepostsandresponses.Intotal,521commentswereprovidedin
responseto18postsofpolls,video,textorphotosforanaverageof28.9commentsperpost.The

9
totalwordcountofallcomments(excludingtheresearcher’sresponsesandcomments)was21,142
words.
Table2
SummaryofFacebookpostsandinteractionwithparticipants.
SurveyQuestion Mainresults(%) Specificcomments
Overall,howmuchdidyoulike
participatingintheFacebook
group?
57.1%Likedsomewhat(N = 12)
38.1%Likedagreatdeal(N = 8)
4.8%Neitherlikednordisliked(N
= 1)
N/A
Didyoufeelthegroupwasasafe
andopenenvironmentforyouto
shareyouropinions?
95.2%Yes(N = 20)
4.8%No(N = 1) No(N = 1):
“Facebookingeneral.I'vetriedto
leaveitafewtimesandthenIalmost
didn'tsignupfor[thisstudy]asit
meantsigningintoFacebook.”(Rae)
Ifyoudidnotparticipateinallof
theposts,whynot? N/A,Iparticipatedinall:57.1%(N
= 12)
Didn’tanswerquestion:14.3%(N
= 3)
Ididn’thavetime:4.8%(N = 1)
Ididn’tnoticethepost:0%
Otherreason:14.3%(N = 3)
Combinationofabovereasons:
9.5%(N = 2)
Otherreasons:
“Busyschedule”(Paige)
“Bookedalastminuteholiday”
(Rosie)
“Ifullyintendtogobackandrespond
tothoseImissed,butsometimesjust
couldn'tgettothemonthedaythey
wereposted.”(Jennifer)
“Bereavementandhospitalwithchild.
AndFacebookbeingprettycrapon
yourphonebrowser.FBwantyouto
usetheirapp.”(Rae)
Onaverage,howdidyoufeel
aboutthelengthofeachpost? Anappropriatelength:95.2%(12)
Toolong:4.8%
Tooshort:0%
N/A
Onaverage,howdidyoufeel
aboutthetimeofdaythatthe
researcherpostedinthegroup?
(Typicallybetween15:30and
17:30onweekdays)
Thetimingofthepostswasfine:
85.7%(N = 18)
Iwishshehadpostedearlierin
theday:14.3%(N = 3)
Iwishshehadpostedlaterinthe
day:0%
N/A
Onaverage,howdidyoufeel
aboutthenumberofpostsinthe
group?(Typically1–2timesper
dayfortwoweeks,with2days
withoutposts1-weekin)
Ithoughtthenumberofpostswas
fine:76.2%(N = 16)
Iwishshehadpostedmoreoften:
14.3%(N = 3)
Iwishshehadpostedlessoften:
9.5%(N = 2)
N/A
Wouldyoutakepartinastudy
likethisagain? Yes:71.4%(N = 15)
Maybe:28.6%(N = 6)
No:0%
Yes(selectedcomments):
“Ifithelpssomeoneelsethenit’s
alwaysworthdoing”(Shannon)
“Itwasanicewaytodothestudy-
wasabletoparticipateatthetime
thatsuitedmeandmyscheduleeach
day.”(Poppy)
“Wasgreattobeabletodipinand
contributewhenhadtime(normally
oncethekidswereinbed)wasgood
tohea[r]otherviewpointsand
experiences”(Rebecca)
“Mademereflectandwasnicetosee
othersinsamepositionreflect.Nice
tothinkitwashelpfulforresearch
too.”(Elizabeth)
“I’malwayshappytotakepartin
studiestohelpresearchintolittle
knownconditionssuchasgd”
(Lorena)
“Yes,it'sgoodtoknowpeoplewant
genuineexperiencestohelpothersin
futureandit'snicetomeetlike-
mindedmums”(Irene)
Maybe(selectedcomments):

10
“It’shardtogettimewhenyouhave
twokidsunder3astheywantallyour
attention.Thenthere’sthedogand
husbandaswellit’shardeventoget
timetodoasurvey.”(Yvonne)
“Thestudywasn'tasdetailedasI
expectedittobe”(Emily)
“ItwilldependsonwhatelseIhave
onduringtheweek.Iwasawayfor
partofthefirstweekandIhadtoplay
catchupforsomeoftheposts”(Allie)
Themostpopularposts(withthehighestnumberofcomments)werethetwopostscontaininga
seriesofimageslistingbarriersandfacilitatorstophysicalactivity–with138and158comments
respectively.Havingconductedpreviousqualitativeinterviewsonfactorsinfluencingphysicalactivity,
theresearcherhadgeneratedalistofbarriersandfacilitatorstophysicalactivityandgroupedthem
accordingtothesocialdeterminantsofhealthmodel(21).Participantswereaskedtochoose1–3
factorswithineachphotoseriesthattheythoughthadthegreatestinfluenceontheirphysicalactivity
levels.Forexample,forfamily-relatedfacilitatorstophysicalactivity(Fig.1),aparticipant
commented,“Iwanttosetagoodexampleformydaughter,andbearoundinthefutureforher.I
wasoutandaboutwithherregularlyintheslingonmaternityleave,notnearlysomuchnowI’mback
atwork”(Mia).Themajorityofotherparticipantsagreedinthatbeingaroundtoseetheirchildren
growupwithasignificantmotivatingfactor.Thissuggestedthatemphasisingthatparticularbenefit
ofphysicalactivity(e.g.itislife-lengtheningandallowsformorehealthyyearsoflifetospendwith
childrenandfamily)mightresonatewiththispopulationgreaterthanotherfacilitators.Fromtallying
thevotesforthemostimportantfactorsfromtheseposts(Fig.2),itallowedtheresearcherto
generatea‘top10’listforbarriersandfacilitators,whichwascriticalinunderstandingwhatto
addressinthetheoryofchange.

Feasibilityofthemethod
Fromtheresearcher’sownmemosandreflectionsonthedatacollection,severallessonsandthemes
werepickedoutoftheonlineworkshop:Pollswereaneasyandpopularwayofgainingconsensus
aboutinterventionoptions.Theywereefficientandwell-answered,withaminimumof18votesper
poll(averageof25votesperpoll,maximumof39votes).Theresearcherallowedforparticipantsto
selectmultipleoptionsaswellasaddtheirownoptiontothepollandletothersvoteonit,which

11
alignedwiththeco-productiongoalsoftheresearch.
Increasingresponserates
SendingoutgentleremindersoverFacebookmessengertoparticipantswhohavenotparticipatedin
theprevious3dayswasausefulwayofboostingparticipationrates.Askingforvariedmethodsof
inputbecameausefultoolingatheringopinionswhenthegroupsloweddownslightlyinthefinal
daysoftheworkshop.DuringDay12,theresearcherpostedavideoaskingforfeedback,andasked
participantstoeithercommentwiththeirthoughtsorto‘react’inacertainwaytoshowtheirviews
(e.g.thumbsupmeant‘like’,angryreactionif‘Idon’tthinkit’sforme’,andlaughingreactionif
‘unsure’).Reactingindifferentwaystopostsallowedallcontenttobecomepollinpart,whichallowed
moretime-constrainedparticipantstostillhavetheirvoicesheard.
Itwasusefulfortheresearchertocheckthepageofteninthefewhoursaftertheposttomoderateif
necessary.Attimes,someonewouldpostsomethingnegativeorcontroversial(nothingsignificant
thatwarrantedfurtheraction),butitwashelpfultorespondquicklytoempathiseand/oraskforother
opinionstoensurethegroupstayedontrack.
Attheendoftheworkshop,theresearcherpostedafinalmessagethankingparticipantsfortaking
partandremindingthemtotakethesurvey.However,someparticipantscontinuedpostinginthe
groupaftertheofficialendofthestudy–twosharedlinksrelevanttothetopicoftheworkshop,and
anotheraskedaquestionrelatedtohergestationaldiabetesfollow-upcare.Thissuggeststhatthe
onlineworkshopmayhavepotentialtobeself-sustaining,astheresearcherdidnotexplicitlymention
orencouragepostingaftertheculminationofthestudy.
Theresultsfromtheend-of-workshopsurveyaredetailedinTable3.Inshort,overalltheworkshop
wasviewedpositively.Itwasenjoyedbyparticipants,with20of21participants(95%)sayingthey
likedtheworkshop‘somewhat’or‘agreatdeal’.Thesamepercentagesaidtheyfeltthegroupwasa
safeandopenenvironmenttoshareopinions,withthedissentingviewmentioningherdislikeof
Facebookmoregenerally.Perhapsthemostimportantindicatorofacceptabilitywasasking
participantsiftheywouldtakepartinsomethingsimilaragain,inwhichcaseover70%(N = 15)said
‘Yes’.Participantswhosaidtheywouldtakepartagainmentionedhowtheformatwasvery

12
convenienttofitintotheirdayandwhentheyhadtime;itallowedthemtoreflectontheirown
experiences;andtheylikedhelpingresearchprogressforgestationaldiabetes.Thesixparticipants
whosaidtheymaytakepartagainsaiditwasstilldifficultfindingtimetodoit,itdependedonwhat
elsewasongoingintheirlives,andoneparticipantsuggesteditwasnotasdetailedasshe
anticipatedittobe.Overall,theworkshopwasacceptabletoparticipants.
SurveyQuestion Mainresults(%) Specificcomments
Overall,howmuchdidyoulike
participatingintheFacebookgroup?
57.1%Likedsomewhat(N=12)
38.1%Likedagreatdeal(N=8)
4.8%Neitherlikednordisliked(N=1)
N/A
Didyoufeelthegroupwasasafeand
openenvironmentforyoutoshareyour
opinions?
95.2%Yes(N=20)
4.8%No(N=1)
No(N=1):
“Facebookingeneral.I'vetriedtoleaveitafewtimesandthenIalmostdidn'tsignupfor[thisstudy]
asitmeantsigningintoFacebook.
Ifyoudidnotparticipateinallofthe
posts,whynot?
N/A,Iparticipatedinall:57.1%(N=12)
Didn’tanswerquestion:14.3%(N=3)
Ididn’thavetime:4.8%(N=1)
Ididn’tnoticethepost:0%
Otherreason:14.3%(N=3)
Combinationofabovereasons:9.5%(N=2)
Otherreasons:
“Busyschedule”(Paige)
“Bookedalastminuteholiday”(Rosie)
“IfullyintendtogobackandrespondtothoseImissed,butsometimesjustcouldn'tgettothemon
thedaytheywereposted.”(Jennifer)
“Bereavementandhospitalwithchild.AndFacebookbeingprettycraponyourphonebrowser.FB
wantyoutousetheirapp.”(Rae)
Onaverage,howdidyoufeelaboutthe
lengthofeachpost?
Anappropriatelength:95.2%(12)
Toolong:4.8%
Tooshort:0%
N/A
Onaverage,howdidyoufeelaboutthe
timeofdaythattheresearcherpostedin
thegroup?(Typicallybetween15:30and
17:30onweekdays)
Thetimingofthepostswasfine:85.7%(N=18)
Iwishshehadpostedearlierintheday:14.3%(N=3)
Iwishshehadpostedlaterintheday:0%
N/A
Onaverage,howdidyoufeelaboutthe
numberofpostsinthegroup?(Typically
1-2timesperdayfortwoweeks,with2
dayswithoutposts1-weekin)
Ithoughtthenumberofpostswasfine:76.2%(N=16)
Iwishshehadpostedmoreoften:14.3%(N=3)
Iwishshehadpostedlessoften:9.5%(N=2)
N/A
Wouldyoutakepartinastudylikethis
again?
Yes:71.4%(N=15)
Maybe:28.6%(N=6)
No:0%
Yes(selectedcomments):
“Ifithelpssomeoneelsethenit’salwaysworthdoing”(Shannon)
“Itwasanicewaytodothestudy-wasabletoparticipateatthetimethatsuitedmeandmy
scheduleeachday.”(Poppy)
“Wasgreattobeabletodipinandcontributewhenhadtime(normallyoncethekidswereinbed)
wasgoodtohea[r]otherviewpointsandexperiences”(Rebecca)
“Mademereflectandwasnicetoseeothersinsamepositionreflect.Nicetothinkitwashelpfulfor
researchtoo.”(Elizabeth)
“I’malwayshappytotakepartinstudiestohelpresearchintolittleknownconditionssuchasgd”
(Lorena)
“Yes,it'sgoodtoknowpeoplewantgenuineexperiencestohelpothersinfutureandit'snicetomeet
like-mindedmums”(Irene)
Maybe(selectedcomments):
“It’shardtogettimewhenyouhavetwokidsunder3astheywantallyourattention.Thenthere’s
thedogandhusbandaswellit’shardeventogettimetodoasurvey.”(Yvonne)
“Thestudywasn'tasdetailedasIexpectedittobe”(Emily)
“ItwilldependsonwhatelseIhaveonduringtheweek.IwasawayforpartofthefirstweekandI
hadtoplaycatchupforsomeoftheposts”(Allie)
Table3.Summarytableoftheend-of-workshopsurveyresults.
Discussion
Theresultsfromthisresearchsuggestthatco-producinganinterventionoverasecretFacebook
groupisasuccessfulapproachtocollectingconsensusandgeneratingideastocreateaphysical
activityinterventionforwomenwithahistoryofgestationaldiabetes.Therewasclearreachand
engagementwiththecontent,asapproximately28commentsperpostweregeneratedby
participants.ThevolumeofthisdatasuggeststhatholdingaworkshopinasecretFacebookgroup

13
wasafeasiblemethodofdatacollectionandtheresultsoftheend-of-workshopsurveysuggestitwas
acceptabletoparticipants.
Theonlineworkshopprovedaveryusefulmethodtodevelopanuancedunderstandingoftheissues
overthetimeperiod.Itdidbecomeclearastheworkshopprogressedwhatthekeyissueswere
regardingphysicalinactivityinthisgroup.Fromdevelopingthetheoryofchange,thetheoryofaction
quicklyfollowedandtheresearcherwasabletoquicklygatherconsensusaboutintervention
componentsbypostinginthesecretFacebookgroup.Theinterventioncontentiscurrentlyunder
developmentandwillbetestedfurtherinlate2020.
Astrengthofthestudywasitsloweconomiccostforresearcherandparticipant.Theonlyresearch
costscamefromresearchertimeand£15honorariumperparticipantforcompletingthefinalsurvey.
TherealsowerenotravelortranscribingcostsfortheFacebookgroup,ascomparedtoatypicalfocus
grouporinterviews.
Anotherkeystrengthoftheresearchwasthatlocationwasnotanobstacletoparticipation.
ParticipantsinthisstudywerespreadthroughoutScotland–geographicspreadandruralityof
participantsistypicallyasignificantbarriertoco-production.Holdinganinterventiondevelopment
workshoponlinevastlyincreasestheinputfromgroupswhomaynotbeabletoattendface-to-face,
duetotime,locationorcircumstantialconstraints.Thisopensupthepossibilityofdoingresearchwith
hardtoreachgroups.
Afurtherstrengthofthismethodwasthat,althoughpotentiallynovelinitsuse,itremainedevidence-
basedintheasynchronousremotecommunity(ARC)literature.Previousresearchhassuggested
optimalmethodsofrecruitment,content,samplesizeandanalysismethod,whichthisstudyfollowed
closely(5,9,10).Additionally,giventhatthismethodallowedforparticipantstotakeasmuchtime
asneededtoconsiderandrespondtoaprompt–incontrasttotheimmediacyofin-personinterviews
andfocusgroups–anotherstrengthcouldbeincreasedthoughtfulnessofanswers.Theresearchers
couldalsotaketimetoconsiderresponsesandaskfollow-upquestionsinamoreconsideredmanner
thanwhatmayoccurduringin-personqualitativeresearch.
TherewereafewlimitationsofthisstudymainlyrelatedtotheuseofFacebook.TheFacebook

14
algorithmofwhatcontentusersseeontheirhomefeedcouldhaveinfluencedtheparticipationrates
ofthisstudy.Someparticipantsmentionedintheend-of-workshopsurveythatiftheydidnotcheck
theFacebookgroupdirectlyforafewdaysandwantedtocatch-up,itwasdifficulttofindtheposts
theyhadandhadnotalreadycontributedtowithoutscrollingthroughallofthecomments.Also,if
participantsengagedlesswiththeFacebookgroupforafewdays,theFacebookalgorithmwould
likelyreducetheamounttheysawtheFacebookgroupontheirtimeline–furthersupressingthe
visibilityoftheFacebookgroupcontent.Therearetwopossiblesolutionstothis:1)Aparticipantin
theend-of-workshopsurveymentionedthattoensurethatsheknewwhenshehadcommentedona
post,she‘liked’thepost.Thisallowedhertoseemoreeasilywhatshehadengagedwith.2)Another
optioncouldbeanexternalchecklistthatparticipantsactivelymarktosignifytheirparticipationin
eachpost.
Anotherlimitationcomeswiththepotentialnegativitiesofusingsocialmediaingeneral–peopleare
reportingeffortstoreducetheirsocialmediause,withdescriptionsofitbeing‘time-wasting’and
potentiallyhavinganegativeimpactontheirmentalhealth(22).Whilemostofthetargetpopulation
usesFacebookdaily,somearetryingtolimituse.ByhavingthisworkshopexclusivelyoverFacebook,
wemaybeunintentionallyexcludingparticipantswhodonotwantorhaveaccesstothistypeof
socialmedia.
Futurestudiesshouldexploreifthismethodisfeasibleandacceptabletoothergroupsofthe
populationandforotherresearchtopicareas.Asdiscussed,thisdemographicarealreadyactive
usersofFacebookwhichlikelycontributedtothehighparticipationratesseen.However,theremay
beothergroupsthatusedifferentsocialmediaorarenotusedtousingFacebook.Twitter,Instagram,
YouTubeandpopularnewmediaformssuchasTikTokhavedifferentcapabilitiesandtarget
demographicstoFacebookandcouldbeusefultoolsforco-productionorworkshopsforappropriate
groups.
Conclusion
ThismethodologypapersuggeststhatusingasecretFacebookgroupforrunninganintervention-
developmentworkshopisacceptabletoparticipants,feasiblefortheresearchertoconductand

15
generateshighquality,nuanceddata.Thismethodholdssignificantpromiseinsimilarfuturework
withgeographicallyremotecommunities,thosewhohavedifficultytravellingorlimitedtime,orthose
withrelativelyuncommondiseasesorriskfactors.
Abbreviations
SixStepsinQualityInterventionDevelopment:6SQuID
Declarations
Ethicsapprovalandconsenttoparticipate
EthicalapprovalwasobtainedpriortothecommencementofthestudybytheSchoolofHealthin
SocialScience’sethicalcommitteeattheUniversityofEdinburgh.Participantsprovidedinformed
onlineconsentpriortotheworkshopbeginning.
Consentforpublication
Participant’sconsentedforanonymisedquotesandaggregatedataregardingage,timesincelast
gestationaldiabetesdiagnosistobedescribedinreportsandpublications.
Availabilityofdataandmaterials
Datasharingisnotapplicabletothisarticleasnodatasetsweregeneratedoranalysedduringthe
currentstudy.
Competinginterests
Theauthorsdeclarethattheyhavenocompetinginterests.
Funding
ThismethodologyresearchwasundertakenaspartofaDiabetesUKPhDstudentship.Thefunding
bodyhadnoroleinthedesignofthestudynordatacollection,analysis,interpretationorinwriting
themanuscript.
Authors’contribution
ABdesignedthestudyandcompleteddatacollection,analysisandinterpretationunderthe
supervisionandadviceofRJandAK.ABdraftedtheinitialmanuscriptwithinputfromRJandAK.
Acknowledgements
Notapplicable.
References

16
1.  W ightD,WimbushE,  J e p s o n  R ,  D o iL.Sixstepsinquality  i n t e r v e n t i o n  d evelopment
(6SQuID).JEpidem i o l  C o m m u n  H.2016;70(5):5 2 0 - 5 .
2.  F a cebook.Faceboo k  Q 2  2 0 1 9 Results.2019.
3.  K ü mpelA,Karnowsk i  V ,  K e y l i n g  T .  Newssharinginsoc i a l  m e d i a :  A  reviewofcurrent
researchonnews  s h a r i n g  u s e r s,content,andnetw o r k s .  S M + S.
2015;1(2):2056305115610141.
4.  N a dkarniA,Hofman n  S G .  W h y  d opeopleuseFace b o o k ?  P e r s  IndivDiffer.
2012;52(3):243-9.
5.  L i j a diA,vanSchalkwy k  G .  O n l i n e  F a cebookfocusgrou p  r e s e a r c h ofhard-to-reach
participants.IJQM.2 0 1 5 ; 1 4 ( 5 ) : 1 60940691562 1 3 8 3 .
6.  F a cebook.WhatAre  t h e  P r i v a c y  O ptionsforFaceboo k  G r o u p s ? :  F acebookHelp
Center.2019.
7.  S o cietyBP.Ethicsguide l i n e s  f o r  i n t e r n et-mediatedresea r c h  ( I N F 2 0 6 /04.2017).2017 .
8.  M acLeodH,JelenB,P r a b h a k a r  A ,OehlbergL,SiekK A ,  C o n n e l l y  K ,editors.Lessons
learnedfromcond u c t i n g  g r o u p -basedresearcho n  f a c e b o o k .Proceedingsofthe  2 0 1 6
CHIConferenceEx t e n d e d  A b s t r actsonHumanFa c t o r s  i n  C o m putingSystems;20 1 6 .
9.  A l q assimMY,Kresnye  K C ,  S i e k  K A ,WoltersMK,editors .  F a c e b o o k forSupportversus
FacebookforRes e a r c h :  T h e  C aseofMiscarriage .  E x t e n d e d  A bstractsofthe201 9  C H I
ConferenceonHum a n  F a c t o r sinComputingSyste m s ;  2 0 1 9 :ACM.
10.  K r e s n yeKC,MaestreJF, J e l e n  B ,  A l q a s simMY,WoltersM K ,  S i e k  K A ,  e d itors.Lessons
LearnedfromRes e a r c h  v i a  P r ivateSocialMediaG r o u p s .  C H I  E xtendedAbstracts;
2019.
11.  B e l l a m yL,CasasJ-P,Hing o r a n i  A D ,  W illiamsD.Type2dia b e t e s  m e l l i t u s after
gestationaldiabetes :  a  s y s t e m a ticreviewandmeta - a n a l y s i s .  L a n cet.
2009;373(9677):1773-9.

17
12.  B e h b oudi-GandevaniS, A m i r i  M ,  Y a r a ndiRB,TehraniFR .  T h e  i m p a c tofdiagnostic
criteriaforgestationa l  d i a b e t e s  o n itsprevalence:asy s t e m a t i c  r e v iewandmeta-
analysis.DiabetolM e t a b  S y n d r .  2019;11(1):11.
13.  B a o  W,TobiasDK,Bow e r s  K ,  C h a v a rroJ,VaagA,Grun n e t  L G ,  e t  a l .  PhysicalActivity
andSedentaryBe h a v i o r s  A s s o ciatedWithRiskofP r o g r e s s i o n  F romGestational
DiabetesMellitustoT y p e  2  D i a b e tesMellitus.JAMAInte r n  M e d .  2 0 14;174(7):1047 .
14.  P a r e ntsandsocialmedia :  M o t h e r s  a reespeciallylikelytog i v e  a n d  r e c eivesupport
onsocialmedia[pr e s s  r e l e a s e ] . PewResearchCe n t e r 2 0 1 5 .
15.  U n d e rhillC,OlmstedM.A n  e x p e r i m e ntalcomparisonof c o m p u t e r - m ediatedandface-
to-facefocusgroup s .  S o c  S c i  C o mputRev.2003; 2 1 ( 4 ) : 5 0 6 - 1 2.
16.  L i j a d i  A ,VanSchalkwykG .  “ T h e  i n t e r n a tionalschoolsareno t  s o  i n t e r n a t i o nalafter
all”:Theeducational e x p e r i e n c e s ofThirdCultureKids.  I J S E P .  2 0 1 8 ;6(1):50-61.
17.  G r o t h SW.Honorarium o r  c o e r c i o n :  u seofincentivesfor p a r t i c i p a n t s  i n clinical
research.JNYState  N u r s e s  A s s oc.2010;41(1):1 1 .
18.  W i l s o n RE,GoslingSD,Gr a h a m  L T .  A  reviewofFaceboo k  r e s e a r c h  in thesocial
sciences.PerspectP s y c h o l  S c i .  2 012;7(3):203-20 .
19.  V a r n hagenC,GushtaM ,  D a n i e l s  J ,  P e tersT,ParmarN,L a w  D ,  e t  a l .  Howinformedis
onlineinformedcon s e n t ?  E t h i c s  B ehav.2005;15(1 ) : 3 7 - 4 8 .
20.  B o w enD,KreuterM,S p r i n g  B ,  C o f t a -WoerpelL,Linnan L ,  W e i n e r  D ,  etal.Howwe
designfeasibilitystud i e s .  A m  J  P r e v Med.2009;36(5 ) : 4 5 2 - 7 .
21.  D a h l g renG,WhiteheadM .  P o l i c i e s  a n d strategiestoprom o t e  s o c i a l  e q uityinhealth.
Stockholm:Institutef o r  f u t u r e  s t u d ies.1991;27(1):4 - 4 1 .
22.  B a u m erEP,AdamsP,  K h o v a n s k a y aVD,LiaoTC,Sm it h  M E ,  S c h w andaSosikV,etal. ,
editors.Limiting,leav i n g ,  a n d  ( r e )  lapsing:anexplorat io n  o f  f a c e b o oknon-use
practicesandexpe r i e n c e s .  P r o c eedingsoftheSIGC H I  c o n f e r e n c eonhumanfacto r s  i n

18
computingsystem s ;  2 0 1 3 :  A C M.
Figures
Figure1
Exampleimageoffamily-relatedfacilitatorstophysicalactivity(identifiedduringprevious
interviewstage).Participantswereaskedtocommentontheimageandlist1-3factors
withinthisthemethatweremostimportanttothem

19
Figure2
Therankingofthetop10factorswomensaidmadephysicalactivityeasier.Womenthen
commentsonthispostwiththeirapprovalandanyadditionalcommentstheywantedto
share.