ASSE Foundation Research Promoting Behavior Change Self-monitoring as an intervention among computer users By Nicole Gravina, Yueng-Hsiang Huang, Michelle M. Robertson, Michael F. Blair and John Austin BEHAVIOR-BASED SAFETY TECHNIQUES grew Cut of a research tradition of behavioral science and irect observations of behavior change over time. ‘These techniques have resulted in improved safe practices and fewer occupational injuries (Komaki, Barwick & Scott, 1978; Krause, Seymour & Sloat, 1995), The process typically involves assessing cur- rent behaviors, determining what they should be to ensure safety and health, communicating these behaviors to the affected workers, developing an ob- servation and feedback system to measure perform- ance, training observers, setting goals for safety performance and, finally, observing workers. The ‘observer then provides feedback to the person being ‘observed and, when appropriate, oflers recognition for atiaining safety goals (McSween, 2004). ‘These observations are typically conducted by a ‘worker's peer who provides immediate verbal feed- back to the observed worker. This feedback can either be positive (eg, because the observed worker fol- Nicole Gravina, M.A, is @ doctoral student inthe applied behavior analysis _program at Western Michigan University. Gravina mas named the 2005 ASSE Foundation Liberty Mutual Safety Research Follows She holds an M.A. in IndstrialOrganizational Psychology and isa member of the Society for Industrial ‘and Organizational Psychology, Assacation of Behavior Analysis and the Organizational Behavior Management (OBM) etvork. ‘Yueng-Hsiang Huang, Ph.D, isa research scientist at Liberty Mutual Research Institute fr Safety in Hopkinton, MMA She holds a Ph.D. in IndustialOrgarizational Psychology from Portiand state University. Huang i @ member of the Sacety for Industrial and Organizational Psychology and the American Psychologal Assocation, ‘She i alo a member ofthe ASSE Foundation Reseerch Committe, ‘Michelle M. Robertson, Ph.D., CPE, 2 research sciemtst atthe Liberty Mutual Research Institute for Safety. She had conducted applied research projects in ergonomics, training and management for more than 17 year. Robertson holds @ PhD. in instructional Technology fram the University of Southern California. She is @ Fellow of the Human Factors and Ergonomics Society. ‘Michael F. Blair, MLM, isa technical consultant with Liberty Mutua Insurance Group, He holds a Master of Management Science degre in Manufacturing Engineering from the University of Massachusets, Lowell He i» member of the ‘Human Fectors and Ergonomics Society John Austin, PhD,, san essocate professor of psychology at Western Michigan University. In adition, hei senior consultant with Aubrey Daniels international _and executive director of the OBM Network. Austin holds a BA from the University (of Notre Dame, and an M.S. and Ph.D, both from Florida State University. 30 PROFESSIONAL SAFETY NOVEMBER 2008. wwwatse.ng lowed prescribed behaviors) or corrective (eg, be- ‘cause the observed worker violated prescribed behav iors) Providing feedback (verbal, written or graphic) as wel asthe act of cbserving itself (Alvero & Austin, 200) have been shown to produce safety-aated bbehavior change (Alvero, Bucklin & Austin, 2001) Although behavior-based safety programs have helped to improve safety in various workplace set tings, the peer observation component limits their application in certain settings—such as where people ‘Work alone or in settings where they cannot leave the ‘work area. Research also stggess that peer observa- tions may be aversive to some employees (Rohn, 2004). Additionally, although some studies have sug. gested that it might not be difficult to encourage Employees to change their safety-related behaviors and postures while they are being observed directly (Gravina, Linstrom-Hlazel & Austin, 2007; Sasson & Austin, 2004), these improvements may diminish over time in the absence of direct observations by others (Lebbon, Austin, Van Houten, et al, mantic script in preparation). Therefore, new approsches that wil ule in safe employee practices even when noone is watching need tobe developed and refined. Self-Monitoring as an Intervention Self-monitoring is one approach within an overall behavioral science strategy that can be used to address the difficulty of applying peer observation in certain situations. Self monitoring is defined as “an individ- val recording the occurrences of his or her own target behavior” (Nelson & Hayes, 1981, p.3). Sel-monitor- ing, along with other behavioral safety intervention components, has been demonstrated to change the performance of truck drivers (Hickman & Geller, 2003), bus operators (Olson & Austin, 2001) and the postures of computer users (McCann & Sulzer- ‘Azaroff, 1996). More specifically, McCann and Sulzer- ‘Azaroff found that computer users worked in neutral postures (overall sitting and hand/wrist) more often fier ergonomics training, discrimination training, SelEmontoring and feedback were ipemented Whereas those researchers used self-monitoring recorded at the end of the session, the current study introduced an automated (eg, automatically prompt- ed by and completed using a computer program) self‘monitoring tool, first at minutes, then extended to 15 ‘minutes, in an effort to provide more opportunites to self-monitor throughout the session. Furthermore, ina sila sdy, Givin Loeny and Aus (ian script in preparation) demonstrated that an intensive seltmonforing procedure couped wih sel-monitor- ing accuracy training resulted in participants assum- ing neutral postures substantially more often than ‘when compared to baseline. However, the self moni- {oring schedule implemented in that study was more intensive (every 2 minutes) than what would be con- ‘sidered reasonable in an office setting ‘Office ergonomic intervention field studies sug- ‘gest that providing adjustable workstations, chairs and forearm suppott yields positive effects on body and visual symptoms (Nelson & Silverstein, 1998) ‘Coupling ergonomic training, with these design fee- tures (Robertson & O'Neill, 3003; Amick, Robertson, DeRango, et al, 2003) or ergonomics training alone (Brisson, Montreuil & Punett, 1999) also yielded pos- itive results on influencing work postures and reducing musculoskeletal discomfort and_pain. Integrating the concept of self-monitoring is one possible approach to encouraging and sustaining behaviors learned in the training, ‘The purpose ofthis study was to extend previous selfemonitoring safety research to evaluate whether the technique is equally effective if the self monitor- ‘ng schedule is extended to a more manageable den- sity. This information could be of value to those inierested in designing effective self-monitoring safety interventions in organizations, especialy for employees who work alone or in locations where peer observations are not feasible. ‘The Study uotatial study participants were recruited ugh a local newspaper advertisement. To be Included priya naw be ple have no history of musculoskeletal disorders and be able to perform data entry tasks using a computer Partic- pants signed an informed consent form in accor- ance with the Institutional Review Board and also gave permision to be photographed or videotaped luring the sty; however, they were not specs iy tal the were bing fed Vina Nader emera ‘Three male and five female participants were includ ced in the study, with ages ranging from 19 to 58. Participants were paid $15 per hour and received a $75 bonus atthe conclusion of the study. Participants performed a data entry faskin which they entered simple text into a commonly used word processing application. Each session lasted $0 min- ‘utes, with a 10-minute break between each session. Each participant completed 20 sessions, with three to seven sessions oceucring on each observation day {or a total of 3 to 5 days across a -week period ‘To reduce the possibility that participants would change behaviors if they knew that behavior change was the primary focus ofthe study, they were initially told thatthe study was designed to assess discom- fort survey aciministored at the end of each 30-minute session. The survey asked participants to rate their level of discomfort for each bady part and shade spe- cific areas of discomfort on a front and back picture of the body part. They completed this survey on their own immediately after each 30-minute session At the end of the study, participants were asked what they perceived tobe the study’s purpose. Al of them reported that it was to assess discomfort. In addition, when asked what they thought the pur ppose of the self-monitoring was, four participants reported that they thought it was to help them keep ‘good postures “in mind”; the other four thought it ‘was meant to measure/check their posture. Study Setting The study took place in a simulate office environ- ‘ment that included a height-adjustable desk, comput er, wrist suppor, clocument holder and an ergonomic chair with several adjustable features. Before data ‘were collected, participants completed a 90-minute ergonomic training session describing neutral pos- tures and how to adjust the workstation and chair (Robertson & O'Neill, 2003; Robertson é& Maynard, 2005). Participants also received an informational sheet highlighting key points ofthe neutral postures. The participants’ knowledge and understanding of the ergonomic principles described during the training were evaluated using a written 21-question test administered before and. after the training. Overall, the average score improved from 149 cor- rect responses on the pretraining test to 193 correct responses on the posttraining test. Using a Mest, this ‘was statistically significant at the .001 level, which ‘means that itis extremely likely that the training is responsible forthe participants’ improved ergonom- ic knowledge as measured by the pre/ posttest. Target Postures Participants’ working postures were recorded via a hidden camera to prevent reactivity to observation. After each session, the video was sampled at 60 sec- ond intervals and each paricipan’s posture, was scored by a trained observer to determine whether a sgiven posture met the definitions of neutral posture 25 described in the training. This information was used to estimate the percentage of neutral postures versus nonneutral postures exhibited curing each session. The observer has previously scored postures for published studies and has previous training in observation and ergonomics. Reliability of measurement was evaluated by comparing the primary observer's scores to an expert ergonomist's scores of samples for each par- ticipant. Agreement averaged 88%. In addition, interobserver agreement was collected by a second trained student observer for 26% of observations and it averaged 90%, This is well above the recom- ‘mended 80% agreement considered to be acceptable in observational research (Kazdin, 1982). Some pos- tures were only viewable by the camera on the left side (as described in the posture definitions) and, therefore, no conclusions can be drawn about the effectiveness ofthe intervention for those postures. ofa behavior-based safe- ‘ty process cn limit its application In certain set- tings—such as people who wark alone or in set- tings whore they cannot leave the work area. Self ‘monitoring is en ‘approach that con be used t0 adres this lini: tation, Tie aril prez lets the results ofa study of the application ofthis technique with a group of computer users ‘wwreasseorg NOVEMBER 206 PROFESSIONAL SAFETY 37Self- monitoring is one approach within an overall behavioral science strategy that can be used to address the difficulty of applying peer observations in certain situations. Figure 1 Observations of Neutral Postures Neck vo, beni 2 . 0 i 1 » o dee SNe es 7 o) | ie 3 ao. 1 | a0. Ao oe | im ge Sle ae 100 — CL 3100 B estes A 20 | aa go NA Pe Ina ios Pp i iv — fm Shoo. Seo Sl E ie ra | x gS Bale | ie is BIVVAM, gE t i a Back a ef bal su vo Soa dtee wae =| | “ West poh tela Green nisin Kall in| saedeall Nee am “ rv o x a Ieee a ean wens lV |e Se "Note. Percentage of obserontions score ns neutml for exch posture by phase for each participant. Each data point repress ‘ane 30-minute session ‘The primary neutral postures measured were based on BSR/HFES 100, Human Factors Engineer- ing of Computer Workstation: Standard for Trial Use, and were defined as follows: ‘*Hlead posture. Head is not bent up or down or tumed left or right. ‘*Back posture. Lower and upper back is suppor ed by the chair. Shoulder posture. Left arm is within 2 in. of the torso (right arm could not be viewed by camera from the side). ‘Arm posture. Left arm elbow angle is between 70" and 135° (right arm could not be viewed by the camera from the side). *Arms supported. At least two thirds of the 32 PROFESSIONAL SAFETY NOVEMBER 2008 wwwasse.org lower left arm is resting on the desk, chair arm sup- port and/or wrist support (right arm could not be ‘viewed by the camera from the side). + Wrist posture, Left wrist is not bent up more than 10° ot down more than 30° (right wrist could not be viewed by the camera from the side), Hip posture. The angle between torso and thigh does not exceed 91" Feet position. Both heel and toe are supported by the floor or a footrest. Before the self-monitoring intervention was intro- duced, each participant was observed and postures that were neutral for the lowest percentage of the obser- vations were noted as target postures for improvement for the participant. These postures (maximum of three)‘were then monitored during the iervention Table 1 The Self-Montoring Observations of Neutral Postures A computer program was Percentage of observations in neutral position used to assist the participants a ‘Self-monitoring Self-monitoring with self-monitoring; it pro- Baseline 2 minutes 15 minutes duced an on-screen display very 2 minutes that prompted Participant 101 Neck 215 590 530 the participant to selémonitor Wrists Ba 197 400 his/her posture, This display Participant 102 Neck 0569 33 included information about the Back 0 995 a9 definition of the targeted pos- Feet 163468 409 tures and asked th pariipantPipant 5 Ams supported _ 0 54 98.63 io score whether s/he was Feet 9510 t00 Pie eat Pomtoring brome, Pariipant 108 Neck Ms 367 32 At the beginning of the study, ‘Ants Supported 39. 69 48 these automated prompts were Participant 105 Neck B64 BLL delivered at 2-minute intervals ‘Arms Supported 0 303 45 because previous research has Feet 160185 215 demonstrated that such an Participant 106 Neck v6 = 31 205 interval is effective for increas- ‘Ams supported 0. 0 0 per terial Cane Paiipantto7 Arms suppored 25, 100. os ciptin preparat iipant ec zlinprepealie) Feet 44.9 9.6 100 Before the self-monitoring process was implemented, par- ticipants also received selEmon- itoring accuracy training. As [Note The average percentage of ebserentions sora as neutral or enh posture by phuse for exch participant ‘These were calculated by dividing the numberof obseroations participants were sored as being the neutral postion by the otal number of obseantons for that phase art ofthis training, participants Ferbalzed whether target posture wes neural or not neutral while they performed the typing task. They were then given feedback from the experimenter regarding the accuracy of the self-assessment. “Haining was finished when at least 20 trials had been completed and accuracy forthe last 10 trials was at least 90%. All participants met the accuracy eriteri- (on within 20 tials with the exception of one partici pant who required 23 tials to reach the criterion ‘At this point, selfmonitoring was implemented during the experimental sessions, each lasting 30 ‘minutes, During this initial intervention phase, par- ticipants self_monitored all oftheir target postures every 2 minutes. ‘After cight solf monitoring sessions were com- pleted, the self-monitoring schedule was extended {o LSsminute intervals. Tis was done to determine whether the increase in neutral postures that result- ed from the 2-minute self-monitoring schedule ‘would continue when the schedule was extended to ‘more reasonable time intervals (.e, 2 minutes vs, 15 minutes. Experimental Design “The study used a single-subject (ie, within-sub- ject) multiple baseline design across participants to {evaluate the effec ofthe intervention. Single-subject designs involve taking. frequent measures, ‘whic allows foreach paticpan’ progress fo be evaluated over time and has been recommended for use in this type of research (Tankersley, McGoey, Dalton, et a. 2006). When using the multiple baseline design, the intervention is implemented in a steggered fashion. When the dependent variable changes imme- diately, and only when the intervention is implement- ted, the change can be attributed with confidence to the independent variable rather than an external factor (eg, practice or a common event in time). Results Alough patent esved tnng egading exgonomie work pracices before the intervention ‘ofan, each partispant il assumed a eat wo non. neutral postures daring the experiment Overall, 18 postures were exposed fo the iervenion, For 12 of Them, th pening of tne they wer coed in the sem poate howe asc eee when ede pllencntochgfncedon oem her dina The up nibcctiee qhosveg guia Be fl 2arnute tervals wos mained above base Une forall postures ith some minor varity up or down) when the ached was extended fo 18 tn thes igue 1 dspags grapes for each posture ex posed oth intervention. abet isplays the average Percentage of dnarations scored ae neta or each Pesture by phase for ach partpant "the percentage of tine partcipants were i the neatal posure es elcid by dividing he noe Tero ofoovations prtidpants were scoed os being wwwasecong NOVEMIER 2008 PROFESSIONAL SAFETY 33in a neutral posture by the fotal number of observa- tions for that phase. As noted, these observations were taken every 60 seconds from the video recordings, Discussion ‘The resulis of this study suggest that selémoni- toring after training may be an effective intervention for influencing long-term, sustained safety-related. behavior changes. In this study, similar improve- ‘ment results were obtained during both the 2minute and 15-minute monitoring schedules, sug- gesting the more practical 15-minute schedule could, be used. In fact, perhaps even less frequent monitor- ing could be effective. Participants reported that the 2-minute schedule was not reasonable, but 7 out of 8 stated that the 15-minute schedule seemed more rea- sonable for a computer-related office jab. One limitation of the study is that participants’ self-monitoring and the video could not be matched ‘exactly; therefore, no information regarding sel mon- itoring accuracy can be reported. As noted, another limitation is that the right side of the body could mot bbe observed from the camera's angle. Thousands of ‘observations of participants’ postures were taken, but the resulls only reflect one side ofthe body and, thus, only provide a limited picture of participants’ re sponse to the self-monitoring intervention. In addition, the findings of this study should be interpreted with caution. Although the intervention appeared to be effective at changing behaviors, other factors that relate to overall safety when using & com- puter were not considered within the design of this study. An increased focus on maintaining specific postures could actually lead to more discamuort, as ‘suggested by studies of heightened muscle activation in computer users maintaining static postures, It's also possible that more dynamic movement and/or sitting in comfortable positions as well as other ergonomic practices such as rest breaks and ‘work environment design would be more effective in decreasing discomfort and injuries. In this case, self-monitoring could be implemented to encourage those behaviors as well. Lastly, the findings do not suggest that self-monitoring should replace safety training, Study participants were exposed to train- ing before data were collected and itis possible that ‘without this information the self-monitoring inter vention could have been less effective Conclusion Selfmontoring may be a viable atemative to peer observations ina. Behaviorbased safety roses. Slt monitoring does not rare feedback Or bseaton By oie wh some employees may find uncomfortable, nor does it equine a person to leave the workstation, which may not be feasible in some work environments, In ain, th feche nique can be used fo encourage behavior change rong employees who work alone, Farther reseneh Should examine the concept of se-moritorng and Schelling as t pertains to various work organiza- fon conditions (eg, est breaks, work demands and pacing) and worlenvironment design factors. 3A PROFESSIONAL SAFETY NOVEMBER 2008 worwasseong References Alvers, AME & Austin 208) The ff bring on thebataro othe cbr mal of Atel Bsr ete sears ‘Alvero AM, Buhl Re Austin J. 200, Anche seo tec a anal csc ‘nee aecbock nog sctings al Oger Tatar nego 338 eae “he, eI Roberson, MM; Dekango,K, et 003) ecto rina intention on reg asl eta mpg. pes 28 2700271 Bean Menta ont, tan cqgsroic sng pops on wort: with iden py nts ‘Scone Wok Eton od, 228253 raving, Ne Aus Scroctde Lret tpi The ses of wenoratorng on ste postr pian are ‘rpm hc Snape ‘Gravina, Ne Lindtonr fart De Ast). 2007, The etc of wort changes nd ao ervetosen Saityping pete ance WORK oa of Preto ‘cn Eten, 28308258 ‘Gravina, N, Lowy, At, Using nenive moni tong scarey nag and af eons prone work pasties Mansi i peprton Hickman) & Gar £5 (03) Slnangamen on cnn eating amon choral tuck ses vl ‘Salen Boer Nngenen 0s Kanlin, AE (80) Sli sh ep Mati or clic et op stings Ne York Onord Unive Pes, Komaki Barwick KD. & Sate LI (79, bal eprom ccna tay pang and enoing ste Fevormancs nad manaacring plan ual ype Bp Sot 45 nase Te Seymour KJ. & Sant, CM. (190 Long- tem evahiaon of bhai bed cod for improving Shy eroonce A meters o 7 teroped ne sae Stone sy Sob C8 ition, fy Asan J, Van Hoste Reta. comparison of cearatanal nd sued dts once bales Mane Sciprispepuaton {cCan 1D Sulzer Azarof, (19%). Cumulative tau mma order taaviorl pty preven a er ural oF ‘Te esr Sn $8 PD MeSween TON) he ee! fy pcs: Inping seat acy ea tn hea appa Coed) New ok jh Wy Sons ‘elu NA & Slvr, A. (198), Workplace changes acces feucon in realest Saplne co ‘ovkars Hit Facts, 0), 720, Nelson ILO. Hayes SC (91), Teo explanations for macy nselietonng Bsr Mean 3 Oso. & Ass}. Behavior ned ac working toe he ete fa selmortring pcg the ‘Sfepetamance obs pesto aro gota! Bde Gralengene 10552 ohh: BG) Eaplring the behavior fan of work ‘nonitoring,Unjsied dct cae, Nese Ngan Unveritkanuzoa Roberson NEM O'Neil, M2609). Reducing muscu vale duro: Blea ofsn ice eget Norhpace {hd alan interven, raion rat 9 ego Sey and Ergmenc 3 1S ihdberson MEM. & Maya WS. 2505 jul). Ofc ergo- orig An intron Sees mated Pasa Bee Season, de Aust J(201 Te fs of tang fend tack and pritpanievenent tekevil sty pal Sitti! Baer Spend 20 ‘Mnteley Ml McGee cE, Dalen Dz etl 2005, Single ssbject esearch mcd eatin WORK: 8 nao Prosi Assn 6 Roubini, 262592. Acknowledgments “he autors thank th ASSE Foundation and Liberty Mutual Resear Insitute fr supporteg ths each tough ths ery Mtl Research allonstip, a wall as Margaret Ron, Glen Pans, ‘Angela Garabe, Bara Fiz and Candace Pry fot the aston