Handwashing with soap (HWWS) has long been recognized as important in reducing infectious disease transmission,1 particularly among those most susceptible to such infections. However, triggering and sustaining improved hand hygiene behaviors remains difficult.2,3 That behaviors change when others are present is a basic tenet of behavioral research, yet the role of social influence on handwashing has been largely unexplored in the literature. Primarily, social influence has been treated as a source of bias—reactivity, courtesy bias, observation bias, and the Hawthorne effect are all concepts used in epidemiology to account for the potential social influence introduced through the act of observation or data collection (Table 1). In hygiene research, social influence is primarily operationalized positively in psychosocial terms—social norms, peer pressure, and social desirability all focus on the psychological processes related to how individuals alter behaviors in a manner that adheres to the expectations of those around them (Table 1). Social influence on handwashing has been examined in health-care settings,4–6 with higher rates of handwashing associated with the presence of an observer or colleague at critical moments. Outside of the health-care setting, few studies have aimed to measure effects of social norms and peer influence on handwashing behavior.7–10 Pickering et al.11 found HWWS among Kenyan primary school students to increase by 23% when at least one other student is present. Although studies are encouraging with respect to peer influence as a tool for handwashing promotion, further study is needed.
Common terms used to describe how social influence alters behaviors
|Reactivity||Modifying one’s behavior as a reaction to being observed13|
|The Hawthorne effect||Describes a specific form of reactivity in which an individual changes their behavior due to the awareness of an experiment, study or the presence of a researcher14|
|Observation bias||Bias in an observer’s measurement or interpretation of their observation that results in misclassification or other error15|
|Experimenter bias||Bias in the observer’s results due to preconceived expectations influencing the experimental design or interpretation16 (also known as “expectancy bias” or “observer-expectancy effect”)|
|Courtesy bias||Modifying behaviors or responses to better fit social norms and/or avoid offending others17|
|Social norms||Societal rules dictating acceptable behavior18|
|Peer pressure||Influence exerted by a peer group that compels someone to conform or act in a certain way19|
|Social desirability||Behaving in a manner that is perceived as desirable or acceptable20|
To explore the role of social influence on HWWS in a primary school setting, we completed a secondary analysis of data from a cluster-randomized controlled trial conducted among 20 schools in Rajshahi, Bangladesh. Eligibility, site selection, and data collection and analysis for the main trial have been previously reported12 (trial registration: NCT02703974). In brief, 20 schools were randomly selected and assigned to receive either a nudge-based handwashing intervention or an intensive hand hygiene education intervention. This analysis examines four post-intervention follow-ups at weeks 6–7, 12–13, 18–19, and 24–25. To allow for an in-depth analysis of social influence beyond the boundaries of intervention designation, we combined data from control and intervention groups, although intervention group was still controlled for in analyses. Details of each intervention group can be found in previous publication.12
Data were collected through video cameras (Super Circuits Covert Hidden Outdoor Electrical Box Spy Camera with Built-in DVR Recorder) disguised as electrical boxes and mounted outside of each school latrine area after approval was granted by the local education office and the school principals. Cameras captured children’s behaviors in the public space entering and exiting the latrines and approaching the handwashing station (HWS). Data were recorded in Excel, noting the time, gender of the child, whether one or both hands were washed, the use of soap and water, if the HWS facility had both soap and water available, and whether another child, teacher, or other adult (such as a neighbor or groundskeeper) was present when the child returned from the toileting event. At times, children urinated and/or defecated outside of the latrine facility and in view of the camera. These were recorded as toileting events and included in our analysis, although the video footage was promptly deleted by the data reviewer. To ensure consistent results, two schools from each follow-up round were randomly selected for re-review with an agreement greater than 95% between the first and second review.
Our dependent outcome variable was washing both hands with soap following a known toileting event. Our primary independent variable was social influence, defined as the presence of another person near the HWS following a toileting event. Because of the limited peripheral range of the cameras, we could only record whether someone was in view of the camera and not necessarily in view of the child.
Social influence was first analyzed as a binary variable, comparing one or more persons present when a student returned from a toileting event to no one present or in view of the camera when returning from the toileting event. In the second analysis, social influence was defined as a categorical variable based on the type of person—no one in view of the camera as the reference group and 1) other child(ren), 2) teacher or other adult, or 3) both child and adult present as comparison groups. Data are presented as the prevalence ratio (PR) calculated using robust Poisson regression, adjusted for gender, school size, intervention group, and school-level clustering. A robust Poisson model was chosen as a more stable alternative to a log-binomial model for calculating changes in the probability or incidence of a binary outcome associated with the independent variable of interest.
HWWS prevalence after toileting events was 63% (990/1,561) when others were present at the time of handwashing compared with 48% (384/799) when the child was alone. This translates to a 30% increase in HWWS in our adjusted model (PR 1.30 CI: 1.14–1.47, P < 0.001) (Table 2). When social influence is defined as a categorical variable, the presence of other children (64%, 932/1,467) or the presence of both children and adults (64%, 33/55) was associated with the highest handwashing prevalence, whereas presence of an adult(s), such as a teacher, groundskeeper, or community member, was associated with a smaller increase in HWWS (59%, 23/39). In our adjusted model, this translates to a 30% increase in HWWS when one or more child is at the HWS (PR 1.30 CI: 1.14–1.49, P < 0.001), a 24% increase when both a child and an adult were present (PR 1.24 CI: 1.01–1.52, P = 0.043), and a 23% increase when one or more adults were present (PR 1.23 CI: 1.03–1.47, P = 0.024) compared with when the child was alone (Table 3).
Washing both hands with soap comparing social influence to no social influence after known toileting events for the four combined follow-up collections
|Social influence||% (N)||Adjusted PR*||Confidence interval||P value|
|No one present or in view of the camera||48% (384/799)||Ref||–||–|
|At least one person present||63% (990/1,561)||1.3||1.14–1.47||< 0.001|
PR = prevalence ratio.
Adjusted for gender, school size, and intervention group.
Washing both hands with soap by type of social influence after known toileting events for the four combined follow-up collections
|Type of social influence||% (N)||Adjusted PR||Confidence interval||P value|
|Total||58% (1,374/2,360)||–||–||Wald test: < 0.001|
|No one present||48% (384/799)||Ref||–||–|
|Children||64% (932/1,467)||1.30||1.14–1.49||< 0.001|
|Both child and adult||64% (35/55)||1.24||1.01–1.52||0.043|
PR = prevalence ratio.
* Adjusted for gender, school size, and intervention group.
Social influence was positively and significantly associated with handwashing in our cluster-randomized trial. Similar to the findings of Pickering and others (2013), our study found that HWWS after a toileting event was 30% higher when another person was present. In reviewing camera footage, we noted several instances in which modeling appeared to be an important mechanism by which social influence influenced handwashing behavior, similar to other studies.11 Examples included students reminding others of handwashing by pointing to or leading another student to the HWS, students demonstrating proper handwashing techniques to other students, and older students assisting younger students with handwashing. Instances of modeling were observed at both nudge and hygiene education schools. The effects of role modeling have also been documented in other settings such as health-care facilities, noting that if the attending physician failed to wash their hands, the other physicians on the team were likely to forgo handwashing as well.6
Even in our limited sample, we found significant differences in handwashing based on the type of person present, with a smaller increase in handwashing observed when an adult was present after a toileting event. Our camera footage suggests that students were at times wary of approaching the HWS if teachers or adults were using it or nearby, possibly out of respect or in an effort to promptly return to class. This highlights the important role adults and teachers can play in influencing handwashing behaviors. A student’s respect for teachers and adults may be a powerful motive for behavior change among school-aged children.
Although the use of cameras may have helped reduce reactivity to the presence of an observer, the camera itself is likely to have engendered reactivity, the independent effect of which is difficult to measure. In addition, our assessment of social influence was based on the cameras’ field of view rather than the student’s own field of view. To address this issue and increase our confidence in our measure of social influence, we conducted a sensitivity analysis in which we isolated the two schools where the handwashing infrastructure was in an enclosed space. The results indicated a similar impact on HWWS, although handwashing rates without another person present were much lower (data not shown).
Social influence could be a powerful tool in promoting handwashing in a primary school setting. Our findings suggest that a hygiene promotion intervention that incorporates social norms as a cue to action could have significant potential to encourage behavior change among primary school students. Fostering positive peer pressure and peer support for improved handwashing should become central to efforts to improve handwashing among school-aged children and the impact on behaviors rigorously documented. The positive potential of social influence could also be considered in the design of school sanitation facilities, ensuring that handwashing facilities are placed in spaces visible to other students. However, caution should be exercised in redesigning facilities, as gender-separated latrines and privacy for girls must be maintained. We, therefore, recommend exploration of a user-centered design for both the handwashing facility and the latrine area that enable social forces to act on the handwashing facility while maintaining gender and privacy needs within the latrine area.
Willmott M, Nicholson A, Busse H, MacArthur GJ, Brookes S, Campbell R, 2016. Effectiveness of hand hygiene interventions in reducing illness absence among children in educational settings: a systematic review and meta-analysis. Arch Dis Child 101: 42–50.
Maury E, Moussa N, Lakermi C, Barbut F, Offenstadt G, 2006. Compliance of health care workers to hand hygiene: awareness of being observed is important. Intensive Care Med 32: 2088–2089.
Pan S-C, Tien KL, Hung IC, Lin YJ, Sheng WH, Wang MJ, Chang SC, Kunin CM, Chen YC, 2013. Compliance of health care workers with hand hygiene practices: independent advantages of overt and covert observers. PLoS One 8: e53746.
Muto CA, Sistrom MG, Farr BM, 2000. Hand hygiene rates unaffected by installation of dispensers of a rapidly acting hand antiseptic. Am J Infect Control 28: 273–276.
Judah G, Aunger R, Schmidt WP, Michie S, Granger S, Curtis V, 2009. Experimental pretesting of hand-washing interventions in a natural setting. Am J Public Health 99 (Suppl 2): S405–S411.
Chittleborough CR, Nicholson AL, Basker E, Bell S, Campbell R, 2012. Factors influencing hand washing behaviour in primary schools: process evaluation within a randomized controlled trial. Health Educ Res 27: 1055–1068.
Pickering AJ, Blum AG, Breiman RF, Ram PK, Davis J, 2014. Video surveillance captures student hand hygiene behavior, reactivity to observation, and peer influence in Kenyan primary schools. PLoS One 9: e92571.
Grover E, Hossain MK, Uddin S, Venkatesh M, Ram PK, Dreibelbis R, 2018. Comparing the behavioural impact of a nudge-based handwashing intervention to high-intensity hygiene education: a cluster-randomised trial in rural Bangladesh. Trop Med Int Health 23: 10–25.
Ram PK 2010. Is structured observation a valid technique to measure handwashing behavior? Use of acceleration sensors embedded in soap to assess reactivity to structured observation. Am J Trop Med Hyg 83: 1070–1076.
Hagel S, Reischke J, Kesselmeier M, Winning J, Gastmeier P, Brunkhorst FM, Scherag A, Pletz MW, 2015. Quantifying the Hawthorne effect in hand hygiene compliance through comparing direct observation with automated hand hygiene monitoring. Infect Control Hosp Epidemiol 36: 957–962.
Strickland B, Suben A, 2012. Experimenter philosophy: the problem of experimenter bias in experimental philosophy. Rev Phil Psychol 3: 457–467.
Ercumen A, Arnold BF, Naser AM, Unicomb L, Colford JM Jr., Luby SP, 2017. Potential sources of bias in the use of Escherichia coli to measure waterborne diarrhoea risk in low-income settings. Trop Med Int Health 22: 2–11.
Cumbler E, Castillo L, Satorie L, Ford D, Hagman J, Hodge T, Price L, Wald H, 2013. Culture change in infection control: applying psychological principles to improve hand hygiene. J Nurs Care Qual 28: 304–311.
Haberecht K, Schnuerer I, Gaertner B, John U, Freyer-Adam J, 2015. The stability of social desirability: a latent change analysis. J Pers 83: 404–412.