'I Just Want to Hack Myself to Not Get Distracted': Evaluating Design Interventions for Self-Control on Facebook
Ulrik Lyngs, Kai Lukoff, Petr Slovak, William Seymour, Helena Webb, Marina Jirotka, Jun Zhao, Max Van Kleek, Nigel Shadbolt
‘‘I Just Want to Hack Myself to Not Get Distracted’:Evaluating Design Interventions for Self-Control on Facebook
Ulrik Lyngs , Kai Lukoff , Petr Slovak , William Seymour , Helena Webb ,Marina Jirotka , Jun Zhao , Max Van Kleek , Nigel Shadbolt Department of Computer Science, University of Oxford, UK{ulrik.lyngs, william.seymour, helena.webb, marina.jirotka,jun.zhao, max.van.kleek, nigel.shadbolt}@cs.ox.ac.uk Human Centered Design & Engineering, University of Washington, Seattle, US, [email protected] Department of Informatics, King’s College London, UK, [email protected]
ABSTRACT
Beyond being the world’s largest social network, Facebook isfor many also one of its greatest sources of digital distraction.For students, problematic use has been associated with nega-tive effects on academic achievement and general wellbeing.To understand what strategies could help users regain control,we investigated how simple interventions to the Facebook UIaffect behaviour and perceived control. We assigned 58 uni-versity students to one of three interventions: goal reminders,removed newsfeed, or white background (control). We loggeduse for 6 weeks, applied interventions in the middle weeks,and administered fortnightly surveys. Both goal remindersand removed newsfeed helped participants stay on task andavoid distraction. However, goal reminders were often an-noying, and removing the newsfeed made some fear missingout on information. Our findings point to future interventionssuch as controls for adjusting types and amount of availableinformation, and flexible blocking which matches individualdefinitions of ‘distraction’.
Author Keywords
Facebook; problematic use; self-control; distraction; ICTnon-use; addiction; focus; interruptions
CCS Concepts • Human-centered computing → Empirical studies inHCI;
INTRODUCTION
Research on ‘Problematic Facebook Use’ (PFU) has investi-gated correlations between Facebook use and negative effectson outcomes such as level of academic achievement [36] andsubjective wellbeing [59, 58]. Here, a cross-cutting findingis that negative outcomes are associated with subjective diffi-culty at exerting self-control over use, as well as specific usepatterns including viewing friends’ wide-audience broadcastsrather than receiving targeted communication from strong ties[13, 59].
Permission to make digital or hard copies of part or all of this work for personal orclassroom use is granted without fee provided that copies are not made or distributedfor profit or commercial advantage and that copies bear this notice and the full citationon the first page. Copyrights for third-party components of this work must be honored.For all other uses, contact the owner/author(s).
CHI ’20, April 25–30, 2020, Honolulu, HI, USA.
Copyright is held by the author/owner(s).ACM ISBN 978-1-4503-6708-0/20/04.http://dx.doi.org/10.1145/3313831.3376672
Much of this work has focused on self-control over Facebookuse in student populations [2, 45, 47], with media multitaskingresearch finding that students often give in to use which pro-vides short-term ‘guilty pleasures’ over important, but aversiveacademic tasks [77, 89, 61]. In the present paper, we present amixed-methods study exploring how two interventions to Face-book — goal reminders and removing the newsfeed — affectuniversity students’ patterns of use and perceived control overFacebook use. To triangulate self-report with objective mea-surement, our study combined usage logging with fortnightlysurveys and post-study interviews.We found that both interventions helped participants stay ontask and use Facebook more in line with their intentions. Interms of usage patterns, goal reminders led to less scrolling,fewer and shorter visits, and less time on site, whereas remov-ing the newsfeed led to less scrolling, shorter visits, and lesscontent ’liked’. However, goal reminders were often experi-enced as annoying, and removing the newsfeed made someparticipants fear missing out on information. After the study,participants suggested a range of design solutions to mitigateself-control struggles on Facebook, including controls for fil-tering or removing the newsfeed, reminders of time spent anduse goals, and removing features that drive engagement. Asan exploratory study, this work should be followed by confir-matory studies to assess whether our findings replicate, andhow they may generalise beyond a student population.
RELATED WORKStruggles with Facebook use
Whereas many uses of Facebook offer important benefits, suchas social support, rapid spread of information, or facilitation ofreal-world interactions [79], a substantial amount of researchhas focused on negative aspects [59]. For example, studieshave reported correlations between patterns of Facebook useand lower academic achievement [78, 87], low self-esteem, de-pression and anxiety [52], feelings of isolation and loneliness[2], and general psychological distress [15]. Such ‘Problem-atic Facebook Use’ (PFU) has been studied under variousnames (including ‘Facebook dependence’ [88] and ‘Facebookaddiction’[5]), but a recent review summarised a common def-inition across papers as ‘problematic behaviour characterisedby addictive-like symptoms and/or self-regulation difficultiesrelated to Facebook use leading to negative consequences inpersonal and social life’ [59]. a r X i v : . [ c s . H C ] M a y large number of studies have in turn correlated measuresof PFU with patterns of use and personality traits. Here, re-searchers often distinguish between use that is more ‘active’(creating content and communicating with friends) and usethat is more ‘passive’ (consuming content created by otherswithout actively engaging), with the former being linked tomore positive correlates of subjective wellbeing [14, 26, 30,32] and the latter to more negative [51, 86].Moreover, most studies have found that ‘problematic users’tend to spend more time on Facebook [59], including a recentstudy by researchers at Facebook with direct access to serverlogs: users who experienced their use as problematic (i.e., re-ported negative impact on sleep, relationships, or work/schoolperformance, plus lack of control over use) spent more time onthe platform, especially at night, as well as more time lookingat profiles and less time browsing the newsfeed, and weremore likely to deactivate their accounts [16].Depending on the specific tools and thresholds used for as-sessing use as ‘problematic’, prevalence estimates vary widely,from 3.1% in a representative sample of US users [16] to 47%in a study of Malaysian university students ([41] see also [8,45, 88]). The upper bounds of such estimates suggest that, atleast at a mild levels, it is very common for people to strugglewith using Facebook in accordance with their goals [35]. Thisis supported by studies of multitasking and media use findingthat people very often perceive their use of digital media tobe in conflict with other important goals (61.2% of use oc-currences in an experience sampling study by Reinecke andHofmann [75]), and that Facebook in particular is one of themost common sources of media-induced procrastination [77,89]. Interventions and digital self-control tools
Catering to users struggling with self-control over digital de-vice use, a growing niche exists for ‘digital self-control’ toolson online stores for apps and browser extensions. Such toolspromise to support user self-control through interventions suchas removing distracting element from websites, tracking andvisualising use, or rewarding intended behaviour [57]. In par-ticular, many browser extensions focus on adjusting
Facebook in ways intended to help self-control, for example by removingthe newsfeed [42] or hiding numerical metrics such as likecount [34].No studies have assessed how interventions found in thesetools may alleviate self-control struggles on Facebook. How-ever, some recent studies have investigated how temporarilydeactivating or not logging in to Facebook affect subjectivewellbeing [4, 63, 83, 85]. The findings from these studieshave largely been in agreement, with Allcott et al. [4] thelargest to date: in a study where 580 participants were ran-domly assigned to deactivate their accounts for four weeks,and compared to 1,081 controls, Facebook deactivation in-creased offline activities (including socialising with familyand friends and watching TV) and subjective wellbeing, anddecreased online activity (including other social media thanFacebook). Moreover, Facebook deactivation caused a largeand persistent reduction in Facebook use after the experiment.For many users, however, deactivating or deleting their Face-book account presents too tall a barrier to action for tacklingproblematic use. Most users have more targeted non-use goals than “abstinence”, such as reducing time scrolling the news-feed (but not time posting in a university social group), orreducing time spent on Facebook during final exams (but notduring vacations, cf. [16, 87]). Some existing research simi-larly supports positive effects on wellbeing of targeted non-use,including research on active versus passive social media use[13, 38, 86]. Therefore, investigating interventions found indigital self-control tools for Facebook presents an excitingresearch opportunity, as they represent less extreme measuresthan deactivation that may have positive effects.
OVERVIEW OF STUDY
On this background, we set out to study how two interven-tions found in popular browser extensions for scaffolding self-control on Facebook — specifically, adding goal prompts andreminders and removing the newsfeed — affect patterns of useand perceived control on Facebook among university students.We designed a mixed-methods study that attempted to addresscommon limitations in related studies: • Most studies rely on self-reported Facebook use, whichcomplicates interpretation because self-report often corre-lates poorly with actual use of digital devices [25, 24, 65,80]. Therefore, we combined surveys and interviews withlogging of use, to triangulate subjective self-report and ob-jective measurement. • Nearly all studies, apart from deactivation studies, have usedcross-sectional designs, making it very difficult to interpretcausality [59]. Therefore, we randomly assigned partici-pants to intervention groups and compared an initial base-line to a subsequent intervention as well as post-interventionblock.Our choice of interventions is described in the ‘Pre-study’section below. Based on existing research on self-controlstruggles in relation to Facebook use, our research questionswere as follows: • RQ1 (Amount of use): How do goal reminders (C goal ) orremoving the newsfeed (C no-feed ) impact time spent andvisits made? • RQ2 (Patterns of use): How do goal reminders (C goal ) orremoving the newsfeed (C no-feed ) impact patterns of use(e.g., passive / active)? • RQ3 (Control): How do goal reminders (C goal ) or removingthe newsfeed (C no-feed ) impact perceived control? • RQ4 (Post-intervention effects): Do the effects (RQ1-3) ofgoal reminders (C goal ) or removing the newsfeed (C no-feed )persist after interventions are removed? • RQ5 (Self-reflection): Do the interventions enable partic-ipants to reflect on their struggles with Facebook use inways that might inform the design of more effective inter-ventions?Whereas RQ1-4 follow from the background literature re-viewed, RQ5 was a generative research question pointing to-wards new design solutions. We did not envision participantsbeing ‘vessels of truth’ in relation to which design interven-tions would solve their struggles, but were interested in whatsuggestions the interventions might inspire as design probes.
ETHODS
Study materials, anonymised data, and analysis scripts areavailable via the Open Science Framework at osf.io/qtg7h.
Pre-study
Reviewing Facebook self-control tools
In February 2018, we searched for browser extensions forsupporting self-control on Facebook on the Chrome Web storeand identified 50 such extensions implementing a range ofinterventions (see study materials for the list). Most (36 out of50) let the user remove or alter distracting elements, with morethan half (27/50) specifically hiding the newsfeed (e.g., ‘News-feed Eradicator’ [42] removes the newsfeed and optionallyreplaces it with a motivational quote). Others implementedinterventions such as time limits (e.g., setting a daily limitand prompting the user to stop using Facebook or, like
AutoLogout [53], force closing it when the time has passed), goalreminders (e.g., asking the user what she needs to do on Face-book and subsequently providing reminders,
Focusbook [27])or providing rewards or punishments (e.g., transferring moneyout of one’s bank account if use is above a set limit,
TimewasteTimer [72]).
Selecting interventions to investigate guided by a dual sys-tems model of self-control
To categorise these interventions, we relied on a recent reviewof functionality in digital self-control tools, which groupedtheir main design features into the types block/removal , self-tracking , goal advancement , and reward/punishment , which inturn were mapped to psychological mechanisms in a dual sys-tems model of self-regulation [57]. This model distinguishesbetween behaviour under non-conscious ‘System 1’ control,i.e., when the external environment or internal states triggerhabits or instinctive responses; and behaviour that is underconscious ‘System 2’ control, i.e., when goals, intentions, andrules held in working memory trigger behaviour. For exam-ple, a student might check Facebook as the first thing whenopening his laptop, because this context triggers a habitualcheck-in via System 1 control. Alternatively, the student mightopen Facebook because he has a conscious goal of messaginga friend.According to this model—which we return to in theDiscussion—‘self-control’ is the capacity of conscious System2 control to override System 1 responses when the two arein conflict. For example, one might have a conscious goal tonot check one’s phone at the dinner table and a need to useself-control to suppress one’s checking habit, in order to alignbehaviour with this goal (see [57] for details, cf. [48]). C no-feed For our first experimental condition, C no-feed , wechose removing the newsfeed, because this was by far the mostcommon approach among the extensions reviewed. Viewedthrough the dual systems model lens, removing the news-feed represents a block/removal strategy which scaffolds self-control on Facebook by preventing unwanted System 1 controlfrom being triggered by the newsfeed, and supporting System2 control by preventing distracting information from crowdingout working memory and make the user forget her goal [57]. C goal To compare this to a different strategy, we selected a goal advancement intervention as a second experimental con-dition (C goal ), specifically the one implemented by
Focusbook
Goal reminder (prompt) Goal reminder (reminder)No newsfeed Control
Figure 1. Mockup of study conditions: C goal (adding a goal prompt whenvisiting the site that every few minutes pops up a reminder), C no-feed (removing the newsfeed), and C control (white background). See studymaterials for screenshots. [27], which prompts the user to type in their goal when vis-iting Facebook and then periodically reminds them of thisgoal. According to the dual systems model, this scaffoldsself-control in a way that is distinct from removing the news-feed, namely, by keeping the goals the user wishes to achievepresent in working memory, thereby enabling System 2 con-trol. We chose
Focusbook ’s implementation, because it hadthe largest number of users among the extensions reviewedthat implemented alternatives to block/removal strategies. C control In order to control for ‘demand characteristics’ andplacebo effects [10, 64], we also included a control condi-tion (C control ). In this condition, we changed the backgroundcolour of Facebook from light grey to white, which we didnot hypothesise to have any significant effect on behaviour orperceived control.
Materials
Study conditions
The study conditions are shown in Figure 1.We implemented the interventions as Chrome extensions writ-ten in JavaScript and CSS: during the intervention block, theextension script for C control turned the background colour ofFacebook white. For C no-feed , the extension script hid thewebpage elements containing the newsfeed. For C goal , theextension script was a modified version of
Focusbook (thesource code for which is available on GitHub [27]), wherewe forced safe-for-work-mode (i.e., avoiding foul language inreminders) and altered prompts that expressed disapproval toneutral reminders (e.g., changing “Fine, just tell me why youneeded to open Facebook” to “Tell me why you needed to openFacebook”). The extension prompted the user to type in whythey opened Facebook when they went to the site, and after1-3 minutes popped up a reminder of what they typed, alongwith a snooze button. Until the snooze button was pressed,the banner containing the prompt slowly expanded to take upmore and more screenspace.
Logging of use
Following recent work [87], we used the open-source browserextension ‘Research tool for Online Social Environments’(ROSE) [70, 71] to log Facebook use in the Google Chromebrowser. We used this extension to record usage metrics (e.g.,imestamps when a browser tab with Facebook was brought inand out of focus, number of clicks) and specific interactions(e.g., viewing a profile, liking content). To preserve privacy,the extension gave interactions (e.g., content liked) an anony-mous identifier in stored data without storing any identifyinginformation about the actual content engaged with. The ROSEextension was installed on participants’ laptop in addition tothe extension for their intervention condition.
Surveys/interviews
Opening survey : The opening survey included (i) demo-graphic information, (ii) basic information about participants’use of Facebook (when they got an account, devices they useto access the site, prior use of self-control tools for Facebook),and (iii) two individual difference measures (susceptibility totypes of distraction [60] and a Big Five personality measure[31]).
Repeated surveys : The survey administered after each studyblock included three measures:(i) The Passive and Active Facebook Use Measure (PAUM;[30]), which assesses frequency of activities on Facebook.The measure is factored into the usage dimensions ‘active so-cial’ (items including “Posting status updates”, “Chatting onFB chat”), ‘active non-social’ (e.g., “Creating or RSVPing toevents”, “Tagging photos”), and ‘passive’ (e.g., “Checking tosee what someone is up to”, “Browsing the newsfeed passively(without liking or commenting on anything)”).(ii) The Multidimensional Facebook Intensity Scale [66],which assesses agreement with statements about Facebookuse (e.g., “I feel bad if I don’t check my Facebook daily”)and is factored into the dimensions ‘persistence’, ‘boredom’,‘overuse’ and ‘self-expression’.(iii) The Single-Item Self-Esteem Scale [76], a commonlyused measure of self-esteem in psychological research.In addition, the survey after the intervention block includeditems on whether the changes affected perceived control, orhow participants accessed Facebook on laptop vs smartphone.
Interviews : After the study, we conducted semi-structuredinterviews with all participants. Main topics probed were(i) whether the interventions worked as expected, (ii) howparticipants experienced the interventions (example question:“When [changes in the participant’s condition], what was thatlike?”), (iii) what changes participants might wish to make toFacebook to support their intended use (example question: “Ifyou could build any extension you wanted to change the wayFacebook appears and works to make it work better for you,what might you want to do?”). : Five months after the study, wesent participants an optional brief survey, assessing whether(and if so, how) the study had led to enduring changes in howthey use Facebook.
Recruitment
Participants were recruited from colleges at the University ofOxford, using a combination of mailouts, posters, and Face-book posts. Recruitment materials described the study as astudy on ‘Facebook distraction’, investigating ‘which partsof Facebook distract users, and what might be done about it’.Recruitment targeted non-first year students aged 18-30, who
Figure 2. Flowchart of the study procedure felt they were ‘often distracted by Facebook’. Participationwas compensated with a £20 Amazon gift card.
Procedure
A flowchart of the study procedure is shown in Figure 2.Participants were randomly assigned to conditions. At aninitial meeting, participants filled in the opening survey and in-stalled two extensions on their laptop for the Chrome browser:the ROSE extension for logging use and our extension formodifying Facebook according to their assigned condition.Participants were instructed to use Chrome whenever they ac-cessed Facebook on their laptop throughout the study period,and informed that the extensions would ‘anonymously mea-sure how you spend time on the site’ and ‘may change howFacebook appears at some point during the study period’. Thelogging period lasted six weeks, grouped into three two-weekblocks. By the end of each block, participants were sent asurvey link on Friday at 3pm and a reminder two days later.The first block served as a baseline, with no changes madeto Facebook. In the second block, interventions were appliedfrom Monday 9am (announced with a pop-up the first timeparticipants visited Facebook) to Monday 9am two weeks later.The third block served as a new baseline measurement (post-intervention) with Facebook returned to normal. By the end ofthis block, a pop-up thanked participants for taking part anddirected them to sign up for an interview and debriefing.A subset of participants (n = 11) began the tracking period oneweek later than the others.
Data pre-analysis
Quantitative data : On rare occasions, the ROSE extensiondid not correctly record entries to or exits from Facebook,which resulted in some instances where the calculated durationof active focus on a tab with Facebook was unrealistically long(more than 24 hours in one case). To handle such instances,we excluded visits longer than one hour when analysing visitdurations (144 tab visits out of a total of 120,002). See study materials for the precise data processing workflow fromraw data to reported results. nterview transcription and thematic analysis : Two of theauthors transcribed and conducted thematic analysis of allthe interviews and free-text survey responses. The recordingswere iteratively transcribed and analysed using an open-codingapproach. The authors reviewed transcripts and identifiedemerging codes individually, and regularly discussed emergingcodes.Thematic analysis was conducted in the Dedoose software;quantitative analyses were conducted in R.
RESULTS
58 students (21 male) took part. For 8 participants, the inter-vention failed (on some Windows laptops, security settingsprompted participants to turn the extensions off), and 1 par-ticipant deactivated his Facebook account during the study.Survey and logging data from these participants, as well astheir interview statements about the interventions, were ex-cluded from analysis. In addition, 2 participants deleted theROSE extension before the debriefing - and with it their loggeduse - and for 1 participant the interview recording device failed.This left us with survey data from 49 participants, logging datafrom 47 participants, and interview data from 57 participantsfor analysis. Median interview length was 23m 51s (sd = 5m5s).In the following, we first report general characteristics of par-ticipants and their Facebook use, as well as introductory noteson how interventions were used and perceived. Afterwards,we report results grouped by research question.
Participant characteristics
Participants’ median age was 22.5 (min = 19, max = 38) years.90% had had a Facebook account for six years or longer, andthe median number of Facebook friends was 900 (min = 200,max = 2200). All participants used Facebook on their laptop,and 96% also used it on their smartphone. On smartphone,most (78%) used the Facebook and Messenger apps, 8% usedthe web browser (instead of the Facebook app) plus the Mes-senger app, 6% used only the Messenger app, and 2% (1participant) used only the smartphone’s web browser to accessFacebook.Most participants (71%) had never used digital self-controltools for Facebook. Among those who had, the most com-monly used tools blocked access (7 participants) or removedthe newsfeed (3 participants). 3 participants currently usedsuch tools; one used
Newsfeed Eradicator (which removesthe newsfeed), another used
Self-control (which blocks socialmedia), and the third used an ad blocker (which we did notconsider a self-control tool).
Overall Facebook use
Across all participants and the entire study period, the mediannumber of daily tab visits to Facebook was 23 (min = 5, max =138). The median break length between visits to Facebook was69.5 seconds (min = 11, max = 445). The median of partici-pants’ average amount of daily time spent was approximately21 minutes (min = 4m, max = 2h 56m).Often, a number of successive tab visits was logged within ashort span of time (e.g., if participants switched back and forthbetween active application windows). Following Cheng et al.[16], we calculated the number of ‘sessions’ as the numberof times where the break between two visits to Facebook was p = D a il y t i m e ( hh : mm : ss ) Condition
ControlGoal reminderRemove newsfeed A p = p = B a s e li ne I n t e r v en t i on P o s t − i n t e r v en t i on N u m be r o f v i s i t s B p = p = B a s e li ne I n t e r v en t i on P o s t − i n t e r v en t i on V i s i t l eng t h ( s e cs ) C Figure 3. Time spent and number of visits made to Facebook. Com-paring baseline and intervention, goal reminders were associated withless daily time (A), fewer tab visits (B), and a trend towards shorter vis-its (C). Removing the newsfeed was associated with shorter visits (C).Comparing the post-intervention block to baseline, goal reminders wereassociated with fewer visits, suggesting an enduring effect of the inter-vention. longer than 60 seconds. The median number of daily sessionson Facebook was 11 (min = 1, max = 101).
Intervention use and perceptions
The C goal extension did not record what participants typedwhen prompted for their goal, as we wanted to study effects ofgoal reminders without participants adapting or self-censoringfrom knowing responses might be read by the researchers.However, we asked in the interviews how they had used it.Most said they wrote short, descriptive, but generic notes forwhat they did (“I would type shorthand in for what I was aboutto do, so most of the time I would say something like ‘reply tomessages’ or just ‘messages’ or ‘post something on a group’or something like that”, P4). Some also said they occasionallywrote meaningless or ‘unsavoury’ things when they foundthe goal prompt annoying or disruptive (“I think sometimesI tried to type in, like, not really proper words and it said,‘give me a proper answer’ and I was like ‘dammit’!”, P27).In C no-feed , one participant said the newsfeed occasionallyflashed on screen very briefly before being hidden by ourscript (“sometimes i saw like a millisecond of something andI was like ‘oh that’s interesting, I would like to see that’ butthen it wasn’t there”, P56).In C control , a couple of participants said the white backgroundmade content stand out less on their screen (“white backgrounddefinitely makes it harder to. . . I don’t think it’s easier toread. . . ”, P1). Others, however, found it aesthetically pleasing(“I just liked Facebook more. . . it felt more. . . I mean itfelt more Nordic, it wasn’t grey and boring, it was white andnice. . . ”, P30) and wanted it to persist (“is there a way that Ican keep the background white?”, P15).
RQ1 (Amount of use): How do goal reminders or remov-ing the newsfeed impact time spent and visits made?
The logging data and qualitative data suggested that C goal ledto less time spent and fewer and shorter visits, whereas C no-feed led to shorter visits (Figure 3):Usage logging showed that in C goal , average daily time onFacebook was significantly lower during the intervention blockthan in the baseline (median daily time in baseline: 27m 14s, = P a ss i v e u s e ControlGoal reminderRemove newsfeed A p = p = p =
246 Baseline Intervention Post−intervention P a ss i v e b r o w s i ng B p = A c t i v e b r o w s i ng CB Figure 4. Scores on the Passive and Active Facebook Use Measure bycondition. Comparing the intervention to the baseline block, removingthe newsfeed reduced scores on the ’passive’ dimension (A), as well as(as expected) individual items ’Browsing the newsfeed passively (withoutliking or commenting on anything)’s (B) and ’Browsing the newsfeed ac-tively (liking and commenting on posts, pictures and updates)’ (C). Goalreminders reduced only passive newsfeed browsing (B). Comparing post-intervention and baseline, removing the newsfeed was associated withreduced passive newsfeed browsing post-intervention (B). median in intervention: 15m 5s, p = 0.01, r = 0.63, Wilcoxonsigned rank test); number of daily visits declined (mediannumber of visits in baseline = 29.4, median in intervention =10.6, p = 0.01, r = 0.63, Wilcoxon signed rank test); and therewas a trend towards shorter visits (mean tab visit duration inbaseline = 1m 25s, mean in intervention = 1m 15s, t (14) =1.96, p = 0.07, d = 0.51). In C no-feed , only visit length declinedsignificantly (mean visit length in baseline = 1m 12s, mean inintervention = 56s, t (13) = 2.81, p = 0.01, d = 0.75). Participants’ reports in the surveys and interviews agreed withthe logging data:In C goal , two common themes were that the intervention re-duced amount of time on Facebook on laptop (“yeah i thinkI used it less and when I was using it I wasn’t using it for verylong, like a minute maybe”, P45 interview3 ; “definitely usedit a bit less”, P21 interview ) and that reduced use was partlycaused by the intervention being annoying/stressful (“Thisprogramme made me annoyed thus I would spent [sic] lesstime on Facebook”, P32 survey ; “The changes stressed me toget done with my task and then close facebook”, P40 survey ).In C no-feed , participants had mixed opinions on whether ornot it reduced amount of use . Some felt it reduced their use(“limited overall usage”, P28 survey , “I think I used it less ermfor shorter periods of time” P55 interview ) but others felt it onlychanged their newsfeed use without affecting amount per se(“The lack of newsfeed is welcome . . . Facebook usage on mylaptop has not changed/barely changed”, P27 survey ; “I spent alot of time actually on facebook but messaging other peopleand not just looking through my wall”, P54 interview ). RQ2 (Patterns of use): How do goal reminders or remov-ing the newsfeed impact patterns of use?
The logging, survey, and interview data suggested that bothC goal and C no-feed affected patterns of use: C goal selectivelyreduced passive scrolling of the newsfeed, whereas C no-feed (as expected) reduced all behaviour related to the newsfeed(Figure 4).Thus, usage logging showed that average daily scrolling de-clined by 42% in C goal (comparing intervention to baseline, t (14) = 2.39, p = 0.03, d = 0.62), and by 73% in C no-feed ( t (13)= 4.15, p = 0.001, d = 1.11). Moreover, in C no-feed , the numberof times content was liked declined (median number of likesduring baseline = 16, median during intervention = 7, p =0.002, r = 0.88, Wilcoxon signed rank test).In the surveys , scores on the Passive and Active Facebook UseMeasure dimensions showed that participants in C no-feed hadsubstantially lower scores on ‘passive’ use in the interventionthan in the baseline block ( t (13) = 4.8, p = 0.0003, d = 1.28).We explored effects on more granular elements of Facebookuse by comparing baseline and intervention scores separatelyfor each item of the PAUM. Two items showed significantvariation with condition: “Browsing the newsfeed passively(without liking or commenting on anything)” and “Browsingthe newsfeed actively (liking and commenting on posts, pic-tures and updates)”: in C goal , participants reported less passive,but not active, browsing of the newsfeed during the interven-tion block compared to baseline (Passive browsing: p = 0.03, r = 0.57, Active browsing: p = 1, r = 0.05, Wilcoxon signedrank test). In C no-feed , participants reported less passive aswell as less active newsfeed browsing (Passive browsing: p =0.001, r = 0.89, Active browsing: p = 0.01, r = 0.69, Wilcoxonsigned rank test). Moreover, participants in C no-feed showeda trend towards lower scores on “Commenting (on statuses,wall posts, pictures, etc)” ( p = 0.09, r = 0.46, Wilcoxon signedrank test).The quantitative results were supported by the qualitative data:for participants in both experimental conditions, a recurrenttheme was that the interventions caused decreased brows-ing of the newsfeed (“I did feel very aware when scrollingdown my newsfeed, and cut it down”, P19 goal_survey ; “defi-nitely meant I spent less time scrolling on newsfeed on mylaptop”, P55 no-feed_survey ), and increased use of Facebookfor other, more deliberate purposes (“a big facebook postor whatever not just passively. . . scrolling”, P41 goal_interview ;“messaging other people and not just looking through my wall”,P54 no-feed_interview ).In C goal , participants said the effects were driven by the in-tervention making them search for reasons to justify beingon the site (“Being asked why I was opening Facebook wasreally helpful as it made me question why”, P41 goal_survey ;“less likely to aimlessly browse, as I couldn’t justify it”,P45 goal_survey ). In C no-feed , participants said the lack of a news-feed made them seek out alternative options that were oftenmore productive and deliberate (“procrastination was more Reported effect sizes are Cohen’s d for t-tests [18] and r forWilcoxon signed rank tests [28], computed with the rstatix pack-age for R [44]. Subscripts indicate whether quotes are from survey free text re-sponses or from post-study interviews, and in some cases also showparticipants’ study condition. The reported p-values are not corrected for multiple comparisons— these should be considered exploratory results to be followed-upwith confirmatory studies.
Intervention effect on perceived control C oun t o f r e s pon s e s Condition
ControlGoal reminderRemove newsfeed
Figure 5. Responses from survey question on control included in thesurvey administered by the end of the intervention block: ’During thelast two weeks, did the changes made to Facebook on your laptop affecthow in control you felt over your use of the site on your laptop?’ productive in that I was uhm seeking things out to read orto do that were more intentional, I suppose, and less kind ofmindless which I guess the newsfeed is”, P12 no-feed_interview ).(Changed patterns of use related to perceived control are re-ported below.)
RQ3 (Control): How do goal reminders or removing thenewsfeed impact perceived control?
The qualitative data suggested that C goal and C no-feed sup-ported control in the sense of helping participants avoid un-intended use and staying on task, but at the cost of beingannoying/frustrating (C goal ) or leading to fear of missing out(C no-feed ).When exploring survey responses in the MultidimensionalFacebook Intensity Scale, the only of its four dimensions thatshowed significant differences between the baseline and inter-vention blocks was overuse : scores on this measure trendedtowards decrease during the intervention in all conditions(C control : t (18) = 2.37, p = 0.03, d = 0.54, C no-feed : t (13) =1.99, p = 0.07, d = 0.53, C goal : t(14) = 1.75, p = 0.1, d = 0.45),perhaps suggesting that the study procedure across conditionsmade participants reflect on use.When asked directly in a survey item following the interven-tion block whether the changes made to Facebook made themfeel less or more in control of their use, C control had no im-pact, while participants seemed divided about the impact ofthe experimental conditions (Figure 5).The qualitative data seemed to provide the explanation:In both C goal and C no-feed , it was a strong theme in the surveysand interviews that the interventions helped participants stayon their intended task during use (“used it less for stuff thatI wasn’t intending when I opened it”, P4 goal_interview ; “I’llkind of forget that I’m doing work and start scrolling so itwas useful to not be able to do that”, P47 no-feed_interview ). Asubtheme was that this included making it easier to disen-gage from use (“it’s good to get this reminder of ‘hey youcan get off this thing’ ”, P31 goal_interview ; “it was easier justto log out, just check what I had to and then leave facebook”P54 no-feed_interview ). In C goal , participants said the reason theintervention helped them stay on task was that it helped themsnap out of automatic use , that is, stop themselves whenthey engaged in unintended behaviour (“[the reminder] sortof snaps you out of that trance, you know what I mean?”,P21 interview ). In C no-feed , participants said the intervention stopped unintended behaviours from being triggered inthe first place (“there is nothing here [referring to the news-feed], like ‘what did I want?’, you know, so then I went andcontacted the person or looked at the specific thing that Iwanted, not what I saw and kinda wanted at the moment”,P56 interview ).At the same time, there were downsides to the interventionsin that C goal was frequently annoying or frustrating , espe-cially because it was not sensitive to context (“I use facebookjust to message people and I found this extremely annoyingbecause I need to tell someone something and then this thingcomes up and I’d just get annoyed. . . ” P32 interview ), and that C no-feed led to fear of missing out (“missing out on a lotbecause actually a lot of the ways I interact with people onfacebook is things I see on the newsfeed”, P12 interview ). Cross-device use
When asked if the interventions changed how they used Face-book on smartphone vs laptop, 86% of participants in C goal and57% in C no-feed answered ‘Yes’ (compared to 6% in C control , χ = 65.19, p < 0.001).Unpacking this in the qualitative data, participants in bothexperimental conditions expressed that cross-device accesshelped them manage the interventions’ downsides, whilestill enjoying the positive effects (“if I was scrolling throughthe newsfeed or checking events, then it wouldn’t be annoyingbecause I shouldn’t be doing that on my laptop while I’mworking, and if it was something like sending messages aboutwork, contacting friends and asking for help then I could usemy phone”, P40 goal_interview ; “I could reap the benefits of thenewsfeed but without being sucked into it on two platforms”,P28 no-feed_survey ), and so they sometimes used their smart-phone for activities on Facebook the interventions inter-fered with, but as a more deliberate choice (“you’re work-ing on your laptop, uhm, and then it’s very easy to just clicknew tab, but having to get your phone out. . . ”, P19 goal_interview ;“the time I did spend on my phone was more, like, focusedbecause I was actually looking for things I missed out on onmy laptop”, P55 no-feed_interview ). RQ4 (Post-intervention effects): Do the effects (RQ1-3) ofgoal reminders or removing the newsfeed persist after in-terventions are removed?
Comparing post-intervention to baseline, C goal and C no-feed were associated with some persisting effects, with participantsin C goal engaging in fewer daily visits and some feeling thatthe intervention helped build a habit of more intentional use,and participants in C no-feed engaging in less passive newsfeedbrowsing.Thus, in terms of amount of use , participants in C goal madefewer daily visits post-intervention compared to baseline (me-dian number of daily visits in first baseline = 29.4, median inpost-intervention block = 9.8, p = 0.003, r = 0.72, Wilcoxonsigned rank test).In terms of patterns of use , participants in C no-feed reportedless passive browsing of the newsfeed post-intervention com-pared to baseline ( p = 0.007, r = 0.78, Wilcoxon signed ranktest). In the interviews, some C no-feed participants expressed feeling less attracted by the newsfeed when it returned (“Ifound myself less interested in the newsfeed”, P10 interview ).In terms of perceived control , some participants in C goal saidhe intervention helped them build a persisting habit of askingthemselves what their intention of use was when visitingthe site (“from this week there is a habit being built. . . askingmyself why I’m opening Facebook and that habit’s perpetuatedmore or less to this week”, P34 interview , “I’m still aware everytime I open Facebook, I’m just a bit more aware every time. . .it’s not the reflex anymore now that I’ve had that experiencewhere I have to write everything down”, P1 interview ). RQ5 (Self-reflection): Do the interventions enable partic-ipants to reflect on their struggles in ways that might in-form the design of more effective interventions?
In the interviews, nearly all participants expressed feelingconflicted about Facebook, in that they found it too useful orengrained in their lives to do without, but also an ongoingsource of distraction and self-control struggles. They readilysuggested a range of design solutions to mitigate self-controlstruggles. The extent to which interventions were perceivedas freely chosen was important to how it was received, andparticipants did not trust Facebook to provide solutions.
Struggles with Facebook use
Too useful to do without, but source of distraction and self-control struggles : On the one hand, Facebook provided func-tionality participants could not — or would not — do without,especially messaging, events, groups, and pages. On the other,Facebook was frequently distracting and caused them to wastetime and feel frustrated (“I just want. . . to hack myself to havethe self-control to, like, not get distracted. . . I literally justuse it as distraction”, P42 no-feed ). In particular, participantsstruggled to use Facebook in line with their intentions. Mainaspects included (i) going to the site to do one thing, but thenforgetting this goal (“there is one specific trigger that I need toopen facebook, but because when I open the page immediatelythere is tons of information there, like erm notifications, andyou scroll down endless streaming. . . so very easily I could bedistracted”, P34 goal ), (ii) internal conflict between short-termgratification and longer-term goals (“might find them [videos]funny in the short term but when I think about it in the biggerpicture it is a complete waste of time”, P48 control ), and (iii)using Facebook purely out of habit. In relation to the latter,emotional states, especially boredom, were mentioned as trig-gers of habitual use (“if I’m in that erm not very motivatedstate. . . I’ll literally just find myself opening it, without eventhinking that I’m doing it”, P17 control ). Specific suggestions for design solutions
Four themes emerged in relation to specific design suggestionsfor mitigating these struggles:
Control over the newsfeed : More than half of participants ex-plicitly said the newsfeed did not give them what they wanted,and they desired easy ways to filter it, limit it, or turn it off.Some had tried customising their newsfeeds, but found Face-book’s means of doing so tedious and ineffective (“I browsethrough shit that I don’t want to see and I keep on clickingon ‘I don’t like this’, ‘this is not interesting’ and of course itkeeps on adding new stuff so that doesn’t solve the problembasically”, P51 control ). Solution suggestions included simpleways to filter the newsfeed (“a slider to modify the amount yousee people who are on your newsfeed at different percentiles”,P49 goal , “two different ones, like you could have a ‘friends’ orlike ‘photos’ or something”, P17 control ), reducing the amount of information (“maybe it should be limited to like ten postsand you wouldn’t get another ten until the next hour”, P45 goal ,“if it was instead like blank and then you opt-in to who youactually wanna see on your newsfeed as opposed to opt-out”,P44 no-feed ), or being able to remove it altogether.
Raise awareness of time spent or usage goals : Participantsoften lost track of time spent, or of their usage goals, andwanted reminders that raised awareness. These should beeasily accessible (“you wouldn’t want it to be buried in settings,something that was actively shown to you I think that would beuseful”, P52 control ), and let users judge whether their use wasappropriate (“if I saw like ‘you’ve spent 2 minutes today’, like‘great, I’ve got loads of time that I can waste tomorrow becauseI’ve been good today’ ”, P6 goal ). Participants in C goal said thetiming and intrusiveness should be calibrated differently to thereminders they experienced in the study (“less in-your-face. . .so maybe more, longer intervals and not the expanding thing. . .if I could change it to longer intervals and maybe a bit lessinvasive then I think it would actually help”, P4 goal ). Remove ‘addictive’ features : Participants wished to removeor modify features driving them to use the site. Specific fea-tures mentioned included notifications (“get rid of notifica-tions. . . if I didn’t have things popping up every 30 minuteslike ‘this has happened’ I don’t think I would think aboutFacebook’, P6 goal ), viral videos, and games (“things like gamesuggestions and like all that sort of stuff I would definitely getrid of cause. . . I don’t want to play games . . . ‘stop buggingme’ ”, P55 no-feed ). One interesting suggestion was to be ableto display content as text-only (“limit it to like text-only postswhen you’re working so that you’re not bothered by videosand algorithms and photos”, P45 goal ). Flexible blocking to meet individual definitions of distrac-tion : Participants suggested blocking solutions that couldadapt to the type — or timing — of use they found distracting.Thus, some said blocking access altogether was too inflexibleto be useful (“there are useful uses of Facebook that aren’t justwaste of time. . . a blanket, like, ‘don’t do anything on Face-book’. . . it’s not practical for those people who have to useFacebook”, P41 goal ). Suggestions for useful solutions includedbeing able to block or allow specific functionality within Face-book, block access only during specific times (“sync it with atimetable, like lectures or something”, P45 goal ), or even auto-matically detect if activity is engaged with as a distraction.
Generic solution needs
People differ in what they seek on Facebook and the de-sign solutions they prefer:
Some participants wanted toblock or remove distractions, whereas others preferred lessintrusive solutions, such as goal reminders. Similarly, eventhough most participants were dissatisfied with the newsfeed,some wanted it to prioritise close ties, whereas others wantedit to prioritise pages they follow (“I wouldn’t want to seeanyone’s posts, I would only want to see posts by things Iwanted to follow, whether that’s petitions or science papers”,P20 no-feed ). Interventions can ‘backfire’ if overly intrusive and/or notfreely chosen:
Participants felt interventions could make peo-ple to rebel against them if too intrusive and/or if they did notfeel in charge. In terms of intrusiveness , some felt blockingtools could backfire for this reason (“I feel like most people igure 6. Summary of main findings for RQ1-4. ’L’ = logged usage data,’S’ = quantitative survey data, ’Q’ = qualitative data from surveys andinterviews. For quantitative data, arrows indicate direction of the effectwhen p < .10, and effect sizes are marked with an asterisk when p < .05. in their nature, if you have something restrictive. . . then youkinda want to rebel against it”, P56 no-feed ). In terms of feelingin control , some participants suggested this could change theirreaction to the very same intervention. For example, a partici-pant in C goal felt the goal reminders were too intrusive and ledto resistance (“I got very used to clicking out of it and like, I’mjust gonna stay on just out of spite”, P19 goal ), but thought shewould react differently if she controlled the reminders herself(“it would be a bit different if it was me, if I could actuallywrite the messages. . . I think that’d help me, and knowing itwas me, so it wasn’t anyone else”). Scepticism about design solutions coming from Facebook:
Participants did not trust Facebook to provide effective solu-tions for mitigating self-control struggles, because this wasseen as going against their business interests (”you wonderhow much they’d try to just give people the information thatdoesn’t really reflect badly on them”, P36 control ; “Facebook’sinterest is for people to spend more time on it ’cause thenthey’ll get more ad revenue, so. . . ”, P45 goal ). DISCUSSION
Figure 6 summarises findings from RQ1-4: both C goal andC no-feed reduced unintended Facebook use (RQ3), with thedownside that C goal was often experienced as annoying andC no-feed made some fear missing out on information (cf.“FOMO”, [73]). On amount of use (RQ1), C goal reduced dailytime, number of visits, and visit length, whereas C no-feed re-duced visit length. On patterns of use (RQ2), C goal and C no-feed reduced scrolling and passive newsfeed browsing, and C no-feed in addition reduced active newsfeed browsing and amountof content ’liked’. On post-intervention effects (RQ4), C goal was associated with fewer visits and C no-feed with less passivenewsfeed browsing.In terms of reflections on struggles and solutions (RQ5), par-ticipants felt conflicted because Facebook was a source ofdistraction and self-control struggles but also vital to stayingconnected, i.e., too useful to avoid. They suggested specific de-sign solutions related to control over the newsfeed, remindersof time spent and usage goals, removing ‘addictive’ features,and flexible blocking. Their preferred solutions (as well as theinformation sought on Facebook) differed, however, and theyfelt that solutions might ‘backfire’ if overly intrusive and/ornot freely chosen. We now discuss design implications as wellas some of the limitations and future work.
Evaluating the experimental interventions
Focusing specifically on the ability to use Facebook in linewith one’s conscious intentions —– which is at the very coreof self-control [22] —– which of our two experimental in-terventions is more effective? Goal reminders and removingthe newsfeed represent contrasting, and potentially comple-mentary, strategies. In our study, both strategies had a pos-itive effect on perceived control and a significant effect onbehaviour, with C goal helping people ’snap out’ of unintendedbehaviour and C no-feed preventing unintended behaviours frombeing triggered. While these results suggest that both interven-tions have potential, as an exploratory study with a restrictedsample, further research is needed to draw definitive conclu-sions about robustness, effect sizes, and individual differences.However, contextualising our study within related research inpsychology and HCI can provide some predictions:One possible approach is to apply a dual systems model of self-regulation (as called for in recent HCI research [19, 69, 1, 57]).From this perspective, goal reminders are a ‘System 2’ in-tervention which supports conscious self-control by bringingthe goals into working memory that the user wishes to controlher behaviour in relation to.
Removing the newsfeed is botha ‘System 1’ and ‘System 2’ intervention which preventsunwanted automatic responses from being triggered by thenewsfeed, and supports conscious self-control by preventingattention-grabbing information from crowding out workingmemory and making the user forget her goal.A recent comprehensive review of digital behaviour changeinterventions found that providing information about the con-sequences of behaviour (a System 2 intervention) tends tobe unsuccessful, despite being the most common technique.The authors argued that targeting unconscious habit forma-tion (System 1) should be the focus for interventions that aimat long-term efficacy [69]. Similarly, psychological researchhas found that people who are better at self-control tend todevelop habits that make their intended behaviour more re-liant on automatic processes (System 1) and less on consciousin-the-moment self-control (System 2), and/or reduce their ex-posure to ‘temptations’ in the first place [29, 22, 23]. This maybe because effective System 2 control depends not only onremembering longer-term goals, but also on one’s motivationto exert control relative to those goals, which can fluctuatewith emotional state (cf. participants who said they were morelikely to go on Facebook when bored or unmotivated [9, 40,55]).We therefore expect removing the newsfeed to be more gen-erally effective than goal reminders, because it reduces theamount of potentially distracting information and thus theneed for in-the-moment conscious control. In our study, thequalitative data did suggest that C goal fostered a habit of askingoneself about one’s purpose when visiting Facebook. However,given the above, the likelihood of effective control througha habit of goal awareness should depend on what content isavailable and how that content is perceived: the more ‘engag-ing’ the content, the greater the risk that goal awareness willnot by itself provide sufficient control motivation [9, 82, 55].Goal reminders should therefore exhibit larger variation ineffectiveness, and may be less useful for individuals whosenewsfeeds contain more attention-grabbing content and/orwho struggle more with inhibiting distractions in general. Thiswould align with recent findings that those who find Facebookore valuable are also (somewhat paradoxically) more likelyto find their use problematic [16]. Similarly, prior work sug-gest that blocking off online distractions is more effective forindividual who are more susceptible to social media distrac-tions [60] (cf. [55, 62]). These strategies are, however, notmutually exclusive and can be combined in effective interven-tions, as is already the case in many digital self-control tools(e.g.,
Todobook [6], which removes Facebook’s newsfeed andreplaces it with a to-do list reminding the user of her goals).
Designing future interventions
Broadly, participants’ suggested design solutions related to ei-ther altering the information landscape (by filtering the news-feed, removing features driving engagement, or blocking dis-tracting elements) or raising awareness to help navigationwithin this landscape (by adding reminders of time spent orusage goals). These suggestions could be compared to themany existing interventions on online stores, analysed usinga dual systems or other model, and strategies more likely tobe effective implemented and evaluated. Here, we discuss im-plications of the cross-cutting theme that interventions shouldbe experienced as freely chosen and not overly intrusive toavoid ‘backfiring’ and motivate people to rebel against anintervention instead of being helped by it (cf. [55]).Given that participants preferred different interventions —with some wanting restrictive blocking tools — it is not asolution to only consider, e.g., non-intrusive addition of usercontrols [37]. Rather, designers should keep in mind that theeffectiveness of the exact same restriction or intrusion maydepend on whether it is perceived by the user as self-imposedor externally imposed [11, 81, 12]. An implication is that in-terventions should be carefully framed as being supportive ofthe user’s personal goals (cf. [81, 7]). For example, blockingtools may wish to remind the user why their past self decidedto impose restrictions on their present self [21]. Current ex-amples ‘in the wild’ include browser extensions for websiteblocking that display motivational quotes or task reminderswhen users navigate to distracting sites [57].One exciting avenue for future tools is systems that can learnthe user’s personal definition of distraction and in what con-texts to, e.g., automatically impose or not impose limits. Thiswas suggested by one of our participants, and is being exploredin some HCI research, e.g.
HabitLab , which rotates betweeninterventions to discover what best helps a user limit time onspecific websites [50]. A useful such system in the context ofFacebook would not simply limit time, but rather assist theuser in carrying out their goals, for example by dynamicallyblocking elements such as the newsfeed if the user’s currentgoal is to create an event. Such a hypothetical system couldbe highly useful, but it would be crucial to its success thatits interventions were perceived by the user as being in herown interest. In addition, it would need to really understandthe user to be functional [56], creating a possible trade-offbetween privacy and the ‘fit’ of the intervention. Facebookitself, with its deep knowledge of user behaviour, might be inthe best position to explore this approach, but we note that par-ticipants in our study were deeply sceptical about Facebook’smotivations and did not expect design solutions coming fromFacebook to be ‘on their side’ (cf. [20, 67]).
Limitations and future work
Confounding variables:
A possible criticism is that lessscrolling and shorter visits when removing the newsfeed oc-cur simply because there is nothing to scroll. We note thatremoving the newsfeed did not make scrolling impossible —it remained relevant on all other pages than the home screen— and thus scrolling remains a useful measure. Moreover,reduced time is often an explicit goal for users, and so timespent in the face of reduced content is a relevant outcome.
Lack of cross-device tracking:
We logged Facebook use onlaptop only and did not quantify effects of the interventionson cross-device use. It is important in future work to assesspotential ‘spillover’ effects between devices when applying in-terventions meant to scaffold self-control [49, 54, 46], and sowe encourage follow-up studies to explore how our methodscould be supplemented by, e.g., smartphone logging.
Retrospective self-report:
In the surveys and interviews, par-ticipants retrospectively reported their experience, which issubject to recall biases [43, 74]. As self-control often involvesone’s past self setting goals for one’s future self (e.g., in block-ing tools), retrospective reflection is very informative [56], butit would be interesting in future research to include experiencesampling methods to assess in-the-moment experience [75].
Granular interventions and usage measures:
Standardmeasures of Facebook use were not optimal for assessinggranular interventions on laptop only: most measures considerglobal use, and factor into broad dimensions. For example, wefound the Passive and Active Facebook Use Measure’s overalldimensions too broad to capture the behavioural changes ourinterventions introduced. We flag this as a consideration forfuture study designs.
Sampling:
Our sample size was restricted, to allow us to con-duct interviews with all participants, and further research isrequired to assess whether our exploratory results will repli-cate (ideally in pre-registered studies with minimum samplesize guided by our effect size estimates, cf. [17]). Moreover,our recruitment was restricted to university students. Whereasprevious research suggests that struggles with Facebook useare particularly pronounced in this population, and that findingeffective interventions in this population therefore is important,further research is needed to assess how our findings mightgeneralise. Finally, our recruitment process may have selectedfor participants who were highly motivated to change theiruse of Facebook and/or who used it extensively. Motivation iscentral to self-control [39], but we did not assess this explicitly.Our participants’ baseline levels of logged use, and scores onthe Multidimensional Facebook Intensity Scale, were fairlyaverage compared to previous studies ([68, 87], see supple-mentary analysis on osf.io/qtg7h), future work will benefitfrom explicitly measuring participants’ level of motivation.
CONCLUSION
Imagining what success for digital self-control on Facebookand beyond looks like is not an academic exercise, but a prac-tical and urgent concern as evidenced by the recent hearing on‘Persuasive Technology’ in the US senate [84], and a UK AllParty Parliamentary Group’s call for a ‘duty of care’ to be es-tablished on social media companies [3]. We encourage futureHCI work in this space to assess possible design interventionswith open and transparent research methods, to provide theevidence base needed to assist regulators in moving towards abenevolent future [33].
CKNOWLEDGEMENTS
We thank Felix Epp for assistance with the ROSE extension;Michael Inzlicht and Nick Yeung for feedback on the studydesign; and Nadia Flensted Høgholt for feedback on the studydesign, participant communication, and design of Facebookusage visualisations.
REFERENCES [1] Alexander T. Adams, Jean Costa, Malte F. Jung, andTanzeem Choudhury. 2015. Mindless Computing:Designing Technologies to Subtly Influence Behavior. In
Proceedings of the 2015 ACM International JointConference on Pervasive and Ubiquitous Computing .ACM, 719–730.
DOI: http://dx . doi . org/10 . . DOI: http://dx . doi . org/10 . [3] All Party Parliamentary Group on Social Media andYoung People’s Mental Health and Wellbeing. 2019. . TechnicalReport. UK Parliament.[4] Hunt Allcott, Luca Braghieri, Sarah Eichmeyer, andMatthew Gentzkow. 2019. The Welfare Effects of SocialMedia . Working Paper 25514. National Bureau ofEconomic Research.
DOI: http://dx . doi . org/10 . [5] Cecilie Schou Andreassen, Torbjørn Torsheim,Geir Scott Brunborg, and Staale Pallesen. 2012.Development of a Facebook Addiction Scale. Psychological Reports
DOI: http://dx . doi . org/10 . . . . PR0 . . . [6] Yummy Apps. 2019. Todobook. (May 2019).[7] Albert Bandura. 1982. Self-efficacy mechanism inhuman agency. American Psychologist
37, 2 (1982),122–147.
DOI: http://dx . doi . org/10 . . . . [8] Fanni Bányai, Ágnes Zsila, Orsolya Király, AnikoMaraz, Zsuzsanna Elekes, Mark D. Griffiths,Cecilie Schou Andreassen, and Zsolt Demetrovics.09-Jan-2017. Problematic Social Media Use: Resultsfrom a Large-Scale Nationally RepresentativeAdolescent Sample. PLOS ONE
12, 1 (09-Jan-2017),e0169839.
DOI: http://dx . doi . org/10 . . pone . [9] Elliot T Berkman, Cendri A Hutcherson, Jordan LLivingston, Lauren E Kahn, and Michael Inzlicht. 2017.Self-Control as Value-Based Choice. Current Directionsin Psychological Science
26, 5 (2017), 422–428.
DOI: http://dx . doi . org/10 . [10] Walter R. Boot, Daniel J. Simons, Cary Stothart, andCassie Stutts. 2013. The Pervasive Problem withPlacebos in Psychology. Perspectives on PsychologicalScience
8, 4 (jul 2013), 445–454.
DOI: http://dx . doi . org/10 . [11] Amara Brook. 2011. Ecological Footprint Feedback:Motivating or Discouraging? Social Influence
6, 2 (April2011), 113–128.
DOI: http://dx . doi . org/10 . . . [12] Gharad Bryan, Dean Karlan, and Scott Nelson. 2010.Commitment Devices. Annual Review of Economics
2, 1(Sept. 2010), 671–698.
DOI: http://dx . doi . org/10 . . economics . . [13] Moira Burke and Robert E. Kraut. 2016. TheRelationship Between Facebook Use and Well-BeingDepends on Communication Type and Tie Strength. Journal of Computer-Mediated Communication
21, 4(2016), 265–281.
DOI: http://dx . doi . org/10 . . [14] Moira Burke, Cameron Marlow, and Thomas Lento.2010. Social Network Activity and Social Well-Being.In Proceedings of the SIGCHI Conference on HumanFactors in Computing Systems (CHI ’10) . ACM, NewYork, NY, USA, 1909–1912.
DOI: http://dx . doi . org/10 . . [15] Wenhong Chen and Kye-Hyoung Lee. 2013. Sharing,Liking, Commenting, and Distressed? The Pathwaybetween Facebook Interaction and PsychologicalDistress. Cyberpsychology, Behavior and SocialNetworking
16, 10 (oct 2013), 728–734.
DOI: http://dx . doi . org/10 . . . [16] Justin Cheng, Moira Burke, and Elena Goetz Davis.2019. Understanding Perceptions of ProblematicFacebook Use: When People Experience Negative LifeImpact and a Lack of Control. In Proceedings of the2019 CHI Conference on Human Factors in ComputingSystems (CHI ’19) . ACM, New York, NY, USA,199:1–199:13.
DOI: http://dx . doi . org/10 . . [17] Andy Cockburn, Carl Gutwin, and Alan Dix. 2018.HARK No More: On the Preregistration of CHIExperiments. In Proceedings of the 2018 CHIConference on Human Factors in Computing Systems(CHI ’18) . ACM, New York, NY, USA, 141:1–141:12.
DOI: http://dx . doi . org/10 . . [18] Jacob Cohen. 1992. A Power Primer. PsychologicalBulletin
DOI: http://dx . doi . org/10 . . . . [19] Anna L Cox, Sandy J J Gould, Marta E Cecchinato,Ioanna Iacovides, and Ian Renfree. 2016. DesignFrictions for Mindful Interactions: The Case forMicroboundaries. In Proceedings of the 2016 CHIConference Extended Abstracts on Human Factors inComputing Systems (CHI EA ’16) . ACM, New York, NY,USA, 1389–1397.
DOI: http://dx . doi . org/10 . .
20] Helen Creswick, Liz Dowthwaite, Ansgar Koene,Elvira Perez Vallejos, Virginia Portillo, Monica Cano,and Christopher Woodard. 2019. “. . . They don’t reallylisten to people”.
Journal of Information,Communication and Ethics in Society
17, 2 (May 2019),167–182.
DOI: http://dx . doi . org/10 . [21] Angela L. Duckworth, Katherine L. Milkman, andDavid Laibson. 2018. Beyond Willpower: Strategies forReducing Failures of Self-Control. PsychologicalScience in the Public Interest
19, 3 (Dec. 2018),102–129.
DOI: http://dx . doi . org/10 . [22] Angela L. Duckworth and Laurence Steinberg. 2015.Unpacking Self-Control. Child developmentperspectives
9, 1 (2015), 32–37.
DOI: http://dx . doi . org/10 . . [23] Angela L. Duckworth, Rachel E. White, Alyssa J.Matteucci, Annie Shearer, and James J. Gross. 2016. AStitch in Time: Strategic Self-Control in High Schooland College Students. Journal of EducationalPsychology
DOI: http://dx . doi . org/10 . [24] David A. Ellis, Brittany I. Davidson, Heather Shaw, andKristoffer Geyer. 2019. Do Smartphone Usage ScalesPredict Behavior? International Journal ofHuman-Computer Studies (may 2019).
DOI: http://dx . doi . org/10 . . ijhcs . . . [25] David A. Ellis, Linda K. Kaye, Thomas D.W.Wilcockson, and Francesca C. Ryding. 2018. DigitalTraces of Behaviour Within Addiction: Response toGriffiths (2017). International Journal of Mental Healthand Addiction
16, 1 (feb 2018), 240–245.
DOI: http://dx . doi . org/10 . [26] Nicole B. Ellison, Charles Steinfield, and Cliff Lampe.2007. The Benefits of Facebook “Friends:” SocialCapital and College Students’ Use of Online SocialNetwork Sites. Journal of Computer-MediatedCommunication
12, 4 (jul 2007), 1143–1168.
DOI: http://dx . doi . org/10 . . . . . x [27] Yonah Forst. 2016. Focusbook. (2016).[28] Catherine O. Fritz, Peter E. Morris, and Jennifer J.Richler. 2012. Effect Size Estimates: Current Use,Calculations, and Interpretation. Journal ofExperimental Psychology. General
DOI: http://dx . doi . org/10 . [29] Brian M. Galla and Angela L. Duckworth. 2015. MoreThan Resisting Temptation: Beneficial Habits Mediatethe Relationship Between Self-Control and Positive LifeOutcomes. Journal of Personality and SocialPsychology
DOI: http://dx . doi . org/10 . [30] Jennifer Gerson, Anke C. Plagnol, and Philip J. Corr.2017. Passive and Active Facebook Use Measure (PAUM): Validation and Relationship to theReinforcement Sensitivity Theory. Personality andIndividual Differences
117 (oct 2017), 81–90.
DOI: http://dx . doi . org/10 . . paid . . . [31] Samuel D. Gosling, Peter J. Rentfrow, and William B.Swann Jr. 2003. A Very Brief Measure of the Big-FivePersonality Domains. Journal of Research in Personality
37, 6 (2003), 504–528.
DOI: http://dx . doi . org/10 . [32] Rachel Grieve, Michaelle Indian, Kate Witteveen, G.Anne Tolan, and Jessica Marrington. 2013. Face-to-Faceor Facebook: Can Social Connectedness Be DerivedOnline? Computers in Human Behavior
29, 3 (may2013), 604–609.
DOI: http://dx . doi . org/10 . . chb . . . [33] Barbara Grimpe, Mark Hartswood, and Marina Jirotka.2014. Towards a Closer Dialogue Between Policy andPractice: Responsible Design in HCI. In Proceedings ofthe SIGCHI Conference on Human Factors inComputing Systems (CHI ’14) . ACM, New York, NY,USA, 2965–2974.
DOI: http://dx . doi . org/10 . . [34] Ben Grosser. 2019. Facebook Demetricator. (jul 2019).[35] Eduardo Guedes, Federica Sancassiani, Mauro GiovaniCarta, Carlos Campos, Sergio Machado, AnnaLucia Spear King, and Antonio Egidio Nardi. 2016.Internet Addiction and Excessive Social Networks Use:What About Facebook? Clinical Practice andEpidemiology in Mental Health : CP & EMH
12 (jun2016), 43–48.
DOI: http://dx . doi . org/10 . [36] Natasha Gupta and Julia D. Irwin. 2016. In-ClassDistractions: The Role of Facebook and the PrimaryLearning Task. Computers in Human Behavior
55 (feb2016), 1165–1178.
DOI: http://dx . doi . org/10 . . chb . . . [37] Jaron Harambam, Dimitrios Bountouridis, MykolaMakhortykh, and Joris van Hoboken. 2019. Designingfor the Better by Taking Users into Account: AQualitative Evaluation of User Control Mechanisms in(News) Recommender Systems. In Proceedings of the13th ACM Conference on Recommender Systems(RecSys ’19) . ACM, New York, NY, USA, 69–77.
DOI: http://dx . doi . org/10 . . [38] Alexis Hiniker, Sungsoo Ray Hong, Tadayoshi Kohno,and Julie A Kientz. 2016. MyTime: Designing andEvaluating an Intervention for Smartphone Non-Use. Proceedings of the 2016 CHI Conference on HumanFactors in Computing Systems (2016), 4746–4757.
DOI: http://dx . doi . org/10 . . [39] Michael Inzlicht, Bruce D. Bartholow, and Jacob B.Hirsh. 2015. Emotional Foundations of CognitiveControl. Trends in Cognitive Sciences
19, 3 (March2015), 126–132.
DOI: http://dx . doi . org/10 . . tics . . .
40] Michael Inzlicht, Brandon J. Schmeichel, and C. NeilMacrae. 2014. Why Self-Control Seems (but May NotBe) Limited.
Trends in Cognitive Sciences
18, 3 (2014),127–133.
DOI: http://dx . doi . org/10 . . tics . . . [41] Hosein Jafarkarimi, Alex Tze Hiang Sim, RobabSaadatdoost, and Jee Mei Hee. 2016. FacebookAddiction among Malaysian Students. InternationalJournal of Information and Education Technology
6, 6(2016), 465–469.
DOI: http://dx . doi . org/10 . . . V6 . [42] JDev. 2019. News Feed Eradicator for Facebook. (jul2019).[43] Daniel Kahneman and Jason Riis. 2005. Living, andThinking about It: Two Perspectives on Life. In TheScience of Well-Being , F. A. Huppert, N. Baylis, andB. Keverne (Eds.). Oxford University Press, New York,285–304.
DOI: http://dx . doi . org/10 . . . [44] Alboukadel Kassambara. 2019. Rstatix: Pipe-FriendlyFramework for Basic Statistical Tests. (dec 2019).[45] Jiraporn Khumsri, Rungmanee Yingyeun, null MereeratManwong, Nitt Hanprathet, and Muthita Phanasathit.2015. Prevalence of Facebook Addiction and RelatedFactors Among Thai High School Students. Journal ofthe Medical Association of Thailand = ChotmaihetThangphaet
98 Suppl 3 (apr 2015), S51–60.[46] Jaejeung Kim, Chiwoo Cho, and Uichin Lee. 2017.Technology Supported Behavior Restriction forMitigating Self-Interruptions in Multi-DeviceEnvironments.
Proceedings of the ACM on Interactive,Mobile, Wearable and Ubiquitous Technologies
1, 3 (sep2017), 1–21.
DOI: http://dx . doi . org/10 . [47] Mustafa Koc and Seval Gulyagci. 2013. FacebookAddiction among Turkish College Students: The Role ofPsychological Health, Demographic, and UsageCharacteristics. Cyberpsychology, Behavior and SocialNetworking
16, 4 (apr 2013), 279–284.
DOI: http://dx . doi . org/10 . . . [48] Hiroki P. Kotabe and Wilhelm Hofmann. 2015. OnIntegrating the Components of Self-Control. Perspectives on Psychological Science
10, 5 (2015),618–638.
DOI: http://dx . doi . org/10 . [49] Geza Kovacs, Drew Mylander Gregory, Zilin Ma,Zhengxuan Wu, Golrokh Emami, Jacob Ray, andMichael S. Bernstein. 2019. Conservation ofProcrastination: Do Productivity Interventions SaveTime Or Just Redistribute It?. In Proceedings of the2019 CHI Conference on Human Factors in ComputingSystems (CHI ’19) . ACM, New York, NY, USA,330:1–330:12.
DOI: http://dx . doi . org/10 . . [50] Geza Kovacs, Zhengxuan Wu, and Michael S. Bernstein.2018. Rotating Online Behavior Change InterventionsIncreases Effectiveness But Also Increases Attrition. Proceedings of the ACM on Human-ComputerInteraction
2, CSCW (Nov. 2018), 1–25.
DOI: http://dx . doi . org/10 . [51] Hanna Krasnova, Helena Wenninger, Thomas Widjaja,and Peter Buxmann. 2013. Envy on Facebook: AHidden Threat to Users’ Life Satisfaction?. In Wirtschaftsinformatik . DOI: http://dx . doi . org/10 . . [52] Leodoro J. Labrague. 2014. Facebook Use andAdolescents’ Emotional States of Depression, Anxiety,and Stress. Health Science Journal
8, 1 (2014), 80–89.[53] AV Tech Labs. 2019. Auto Logout. (jul 2019).[54] Laura Lascau, Priscilla N Y Wong, Duncan P Brumby,and Anna L Cox. 2019. Why Are Cross-DeviceInteractions Important When It Comes To DigitalWellbeing? (2019).[55] Min Kyung Lee, Sara Kiesler, and Jodi Forlizzi. 2011.Mining Behavioral Economics to Design PersuasiveTechnology for Healthy Choices. In
Proceedings of theSIGCHI Conference on Human Factors in ComputingSystems (CHI ’11) . ACM, New York, NY, USA,325–334.
DOI: http://dx . doi . org/10 . . [56] Ulrik Lyngs, Reuben Binns, Max Van Kleek, and NigelShadbolt. 2018. "So, Tell Me What Users Want, WhatThey Really, Really Want!". In Extended Abstracts ofthe 2018 CHI Conference on Human Factors inComputing Systems (CHI EA ’18) . ACM, New York, NY,USA, alt04:1—-alt04:10.
DOI: http://dx . doi . org/10 . . [57] Ulrik Lyngs, Kai Lukoff, Petr Slovak, Reuben Binns,Adam Slack, Michael Inzlicht, Max Van Kleek, andNigel Shadbolt. 2019. Self-Control in Cyberspace :Applying Dual Systems Theory to a Review of DigitalSelf-Control Tools. In CHIConference on HumanFactors in Computing Systems Proceedings (CHI 2019) .ACM, New York, NY, USA.
DOI: http://dx . doi . org/10 . . [58] Claudia Marino, Gianluca Gini, Alessio Vieno, andMarcantonio M. Spada. 2018a. The Associationsbetween Problematic Facebook Use, PsychologicalDistress and Well-Being among Adolescents and YoungAdults: A Systematic Review and Meta-Analysis. Journal of Affective Disorders
DOI: http://dx . doi . org/10 . . jad . . . [59] Claudia Marino, Gianluca Gini, Alessio Vieno, andMarcantonio M Spada. 2018b. A ComprehensiveMeta-Analysis on Problematic Facebook Use. Computers in Human Behavior
83 (2018), 262–277.
DOI: http://dx . doi . org/10 . . chb . . .
60] Gloria Mark, Mary Czerwinski, and Shamsi T Iqbal.2018. Effects of Individual Differences in BlockingWorkplace Distractions. In
Proceedings of the 2018 CHIConference on Human Factors in Computing Systems(CHI ’18) . ACM, New York, NY, USA, 92:1—-92:12.
DOI: http://dx . doi . org/10 . . [61] Adrian Meier, Leonard Reinecke, and Christine E.Meltzer. 2016. Facebocrastination? Predictors of UsingFacebook for Procrastination and Its Effects on StudentsWell-Being. Comput. Hum. Behav.
64, C (nov 2016),65–76.
DOI: http://dx . doi . org/10 . . chb . . . [62] Pardis Miri, Andero Uusberg, Heather Culbertson,Robert Flory, Helen Uusberg, James J. Gross, KeithMarzullo, and Katherine Isbister. 2018. EmotionRegulation in the Wild: Introducing WEHAB SystemArchitecture. In Extended Abstracts of the 2018 CHIConference on Human Factors in Computing Systems(CHI EA ’18) . ACM, New York, NY, USA, ArticleLBW021, 6 pages.
DOI: http://dx . doi . org/10 . . [63] Roberto Mosquera, Mofioluwasademi (Moffii)Odunowo, Trent McNamara, Xiongfei Guo, and RaganPetrie. 2019. The Economic Effects of Facebook . SSRNScholarly Paper ID 3312462. Social Science ResearchNetwork, Rochester, NY.[64] Austin Lee Nichols and Jon K. Maner. 2008. TheGood-Subject Effect: Investigating Participant DemandCharacteristics.
The Journal of General Psychology
DOI: http://dx . doi . org/10 . . . . [65] Amy Orben and Andrew K. Przybylski. 2019. Screens,Teens, and Psychological Well-Being: Evidence FromThree Time-Use-Diary Studies. Psychological Science
30, 5 (may 2019), 682–696.
DOI: http://dx . doi . org/10 . [66] Gábor Orosz, István Tóth-Király, and Beáta B˝othe.2016. Four Facets of Facebook Intensity — TheDevelopment of the Multidimensional FacebookIntensity Scale. Personality and Individual Differences
100 (oct 2016), 95–104.
DOI: http://dx . doi . org/10 . . paid . . . [67] Elvira Perez Vallejos, Ansgar Koene, Virginia Portillo,Liz Dowthwaite, and Monica Cano. 2017. YoungPeople’s Policy Recommendations on AlgorithmFairness. In Proceedings of the 2017 ACM on WebScience Conference (WebSci ’17) . ACM, New York, NY,USA, 247–251.
DOI: http://dx . doi . org/10 . . [68] Becky Phu and Alan J. Gow. 2019. Facebook Use andIts Association with Subjective Happiness andLoneliness. Computers in Human Behavior
92 (mar2019), 151–159.
DOI: http://dx . doi . org/10 . . chb . . . [69] Charlie Pinder, Jo Vermeulen, Benjamin R. Cowan, andRussell Beale. 2018. Digital Behaviour Change Interventions to Break and Form Habits. ACMTransactions on Computer-Human Interaction
25, 3(2018), 1–66.
DOI: http://dx . doi . org/10 . [70] Andreas Poller. 2017. Welcome to ROSE - TheResearch Tool for Online Social Environments. (2017).[71] Andreas Poller, Petra Ilyes, Andreas Kramm, and LauraKocksch. 2014. Investigating OSN Users’ PrivacyStrategies with In-Situ Observation. In Proceedings ofthe Companion Publication of the 17th ACM Conferenceon Computer Supported Cooperative Work & SocialComputing (CSCW Companion ’14) . ACM, New York,NY, USA, 217–220.
DOI: http://dx . doi . org/10 . . [72] Prettymind.co. 2018. Timewaste Timer .[73] Andrew K. Przybylski, Kou Murayama, Cody R.Dehaan, and Valerie Gladwell. 2013. Motivational,Emotional, and Behavioral Correlates of Fear of MissingOut.
Computers in Human Behavior
29, 4 (2013),1841–1848.
DOI: http://dx . doi . org/10 . . chb . . . [74] Donald A. Redelmeier and Daniel Kahneman. 1996.Patients’ Memories of Painful Medical Treatments:Real-Time and Retrospective Evaluations of TwoMinimally Invasive Procedures. Pain
66, 1 (1996), 3–8.
DOI: http://dx . doi . org/10 . [75] Leonard Reinecke and Wilhelm Hofmann. 2016.Slacking off or Winding down? An ExperienceSampling Study on the Drivers and Consequences ofMedia Use for Recovery versus Procrastination. HumanCommunication Research
42, 3 (jul 2016), 441–461.
DOI: http://dx . doi . org/10 . . [76] Richard W. Robins, Holly M. Hendin, and Kali H.Trzesniewski. 2001. Measuring Global Self-Esteem:Construct Validation of a Single-Item Measure and theRosenberg Self-Esteem Scale. Personality and SocialPsychology Bulletin
27, 2 (2001), 151–161.
DOI: http://dx . doi . org/10 . [77] Larry D. Rosen, L. Mark Carrier, and Nancy A. Cheever.2013. Facebook and Texting Made Me Do It:Media-Induced Task-Switching While Studying. Computers in Human Behavior
29, 3 (2013), 948–958.
DOI: http://dx . doi . org/10 . . chb . . . [78] Sana Rouis, Moez Limayem, and Esmail Salehi-Sangari.2011. Impact of Facebook Usage on Students’Academic Achievement: Role of Self-Regulation andTrust. Electronic Journal of Research in EducationalPsychology
25 (2011), 34.[79] Tracii Ryan, Andrea Chester, John Reece, and SophiaXenos. 2014. The Uses and Abuses of Facebook: AReview of Facebook Addiction.
Journal of BehaviouralAddictions
3, 3 (2014), 133–148.
DOI: http://dx . doi . org/10 . . . .
80] Michael Scharkow. 2016. The Accuracy ofSelf-Reported Internet Use - A Validation Study UsingClient Log Data.
Communication Methods andMeasures
10, 1 (jan 2016), 13–27.
DOI: http://dx . doi . org/10 . . . [81] Janet K. Swim and Brittany Bloodhart. 2013.Admonishment and Praise: Interpersonal Mechanismsfor Promoting Proenvironmental Behavior. Ecopsychology
5, 1 (March 2013), 24–35.
DOI: http://dx . doi . org/10 . . . [82] D M Tice, E Bratslavsky, and R F Baumeister. 2001.Emotional Distress Regulation Takes Precedence overImpulse Control: If You Feel Bad, Do It! Journal ofpersonality and social psychology
80, 1 (2001), 53–67.
DOI: http://dx . doi . org/10 . . . . [83] Morten Tromholt. 2016. The Facebook Experiment:Quitting Facebook Leads to Higher Levels ofWell-Being. Cyberpsychology, Behavior and SocialNetworking
19, 11 (nov 2016), 661–666.
DOI: http://dx . doi . org/10 . . . [84] U.S. Senate Committee On Commerce, Science, &Transportation. 2019. Optimizing for Engagement:Understanding the Use of Persuasive Technology onInternet Platforms. (jun 2019). . commerce . senate . gov/public/index . cfm/2019/6/optimizing-for-engagement-understanding-the-use-of-persuasive-technology-on-internet-platforms [85] Eric J. Vanman, Rosemary Baker, and Stephanie J.Tobin. 2018. The Burden of Online Friends: The Effects of Giving up Facebook on Stress and Well-Being. TheJournal of Social Psychology
DOI: http://dx . doi . org/10 . . . [86] Philippe Verduyn, David Seungjae Lee, Jiyoung Park,Holly Shablack, Ariana Orvell, Joseph Bayer, OscarYbarra, John Jonides, and Ethan Kross. 2015. PassiveFacebook Usage Undermines Affective Well-Being:Experimental and Longitudinal Evidence. Journal ofExperimental Psychology. General
DOI: http://dx . doi . org/10 . [87] Yiran Wang and Gloria Mark. 2018. The Context ofCollege Students’ Facebook Use and AcademicPerformance: An Empirical Study. In Proceedings of the2018 CHI Conference on Human Factors in ComputingSystems (CHI ’18) . ACM, New York, NY, USA,418:1–418:11.
DOI: http://dx . doi . org/10 . . [88] Isabella Wolniczak, José Alonso Cáceres-DelAguila,Gabriela Palma-Ardiles, Karen J. Arroyo, RodrigoSolís-Visscher, Stephania Paredes-Yauri, KarinaMego-Aquije, and Antonio Bernabe-Ortiz. 2013.Association between Facebook Dependence and PoorSleep Quality: A Study in a Sample of UndergraduateStudents in Peru. PloS One
8, 3 (2013), e59087.
DOI: http://dx . doi . org/10 . . pone . [89] Shan Xu, Zheng (Joyce) Wang, and Prabu David. 2016.Media Multitasking and Well-Being of UniversityStudents. Comput. Hum. Behav.
55, PA (feb 2016),242–250.
DOI: http://dx . doi . org/10 . . chb . . ..