Disciplinary behaviour management strategies  in schools and their impact on student psychosocial outcomes: A systematic review

Objectives

To examine whether the use of disciplinary behaviour management strategies (interventions) in secondary schools leads to adverse psychosocial outcomes for pupils

Secondary objectives were:

This review was registered with the international Prospective Register of Systematic Reviews (PROSPERO) in October 2021. Registration number CRD42021285427²¹.

Public Involvement

We held three involvement sessions with young people aged between 11 and 16 years old to develop the broad research questions. The first session involved the Young People’s Advisory Group (YPAG; a local public involvement group for young people interested in research) raising concerns about the effect of disciplinary behaviour management strategies on pupil wellbeing. A second workshop was conducted with funding from Create to Collaborate²² to explore these concerns with a broader group of young people affiliated with a mental health charity. In this workshop, young people suggested that some school discipline practices affect their wellbeing negatively. We then ran a third workshop to refine the review questions and search terms with the input from YPAG.

Eligibility criteria

We included randomised and non-randomised study designs (including longitudinal and cross-sectional surveys). We excluded solely qualitative studies as this review’s scope was limited to effects of interventions.

Based on our public involvement work with young people, we were interested in zero-tolerance, punishment-based, or punitive disciplinary strategies that include verbal reprimanding, behaviour monitoring and reporting, isolation, detentions (either during- or after- school hours), and suspension (inclusive of temporary- or fixed- term exclusion). We did not include studies which only focused on permanent exclusion or expulsions from schools.

We limited our inclusions to the UK and other high-income countries in The Organisation for Economic Co-operation and Development (OECD). This meant that approaches such as corporal punishment, and physical- or chemical- restraint were not included given that they are not implemented in a UK (or similar) context.

We included studies of children and young people from the general population, aged 11–16 years, attending a main-stream school. We excluded studies focused on pupils in specialist schools, such as secure centres for children (similar to a juvenile correction facility in the USA), special behavioural units, and Special Educational Needs and Disability (SEND) schools.

Our primary outcomes were any measures of mental health and wellbeing. We included academic and social outcomes as secondary outcomes.

Search strategy

We developed search strategies with an information specialist (SD) and searched seven online databases from inception to October 15th, 2021: MEDLINE; Embase; PsycINFO; British Education Index; Australian Education Index; Education Resources Information Centre (ERIC); Web of Science Social Science Citation Index (SSCI). See Extended data for search strategies²³. We also sent a standardised email through the Children and Young People’s Mental Health Coalition²⁴ to their 237 member organisations to help identify grey literature.

Study selection

Titles and abstracts identified through electronic database- and web- searching were independently screened for relevance in duplicate (JN, SI, & LM) using Rayyan²⁵. Full texts were then retrieved for all relevant references and assessed against the inclusion criteria, in duplicate. Reasons for exclusion were documented (see table S1 in supplementary file) at this stage. Any discrepancies between reviewers at either stage were resolved through discussion or via a third reviewer (JS).

Data extraction and risk of bias assessment

Data extraction was undertaken by one reviewer (JN) using a standardised form in Microsoft Excel. To minimise bias and errors, a second reviewer (SI) checked the data extracted from all included papers. We extracted information on the following: a) study design, b) sample size and characteristics, c) the behaviour management strategy being studied, d) control / comparator [where available], e) context and setting, and f) information about, and results pertaining to, the primary and secondary outcomes. We assessed risk of bias in included studies using Cochrane Effective Practice and Organisation of Care (EPOC) group’s criteria for nonrandomised studies²⁶. We considered a study to be at an overall low risk of bias when all items were scored at low risk, at an overall moderate risk of bias when more than half the items were at low risk of bias, and all others were rated high risk of bias.

Synthesis

We planned for a random effects meta-analysis if combinable data were available. However, such data across studies were not available and a narrative synthesis was therefore performed.

Study selection

Electronic searches resulted in 5357 citations. We found no additional studies through contacts with experts and third sector organisations. Fifty papers were included for full text assessment. After full text screening, 14 studies^27–40 out of these 50 were included in narrative synthesis. See PRISMA Flow Diagram in extended data²³.

Description of included studies

Ten studies were from USA, followed by two from Australia and the UK respectively, and one from Poland. We found no randomised trials. Studies were either surveys or uncontrolled before and after designs. Sample sizes across studies varied widely, ranging from 23 pupils to 33 572 pupils (median = 1811 pupils).

Studies typically included more females than males. Most studies (n=9)^{27,28,30,31,33,35–38} included high school pupils. White pupils were included less often (mean 36%, median 30% across studies) than non-white pupils. Five studies^{29,33,37,39,40} focused on deprived populations, reported as majority (>50%) children being on free school meals and/or low earning.

Temporary suspension from school was most frequently studied (n=11)^{27,29,31–33,35,37–41}, followed by verbal reprimanding or punishment (n=2)^34,36 and various mixed (multiple combined) strategies (n=2)^28,30. We present result for these categories separately below.

The most common studied outcome was depression (n=7)^{28–30,32,33,35,38} using various scales: PHQ-9 (Patient health Questionnaire); CESD (Center for Epidemiological Studies Depression); SDQ (Strengths and Difficulties Questionnaire); California Healthy Kids Survey (WestEd) (Depression subscale); and Add Health survey. Two studies assessed internalising symptoms (i.e., problems of withdrawal, somatic complaints, and anxiety/depression)⁴², one using Teacher observation of classroom adaptation checklist (TOCAC)²⁹ and Youth Self Report (YSR) in the other³⁹. One study assessed externalising symptoms (i.e., problems of aggression, impulsivity, and inattention)⁴² on adapted Behavior Assessment System for Children: second edition (BASC 2) scale²⁷. Anxiety was assessed in one study³⁴ using a Polish version of State-Trait Anxiety Inventory for Children. General mental wellbeing was assessed in three studies^31,36,37, one using the Warwick and Edinburgh Mental Wellbeing Scale³¹, and two using author developed scales^36,37. Five studies^{28,33,34,37,40} assessed impact of disciplinary strategies on educational attainment. See Table 1 for details of included studies.

Risk of bias in included studies

Studies were mostly at high risk of bias across all domains (see Figure 1). Only one study²⁹ was considered at an overall moderate risk of bias. None were considered at a low risk of bias. For most studies there is risk of bias due to confounding. For cross-sectional studies and surveys the risk of reverse causality is a key problem, i.e., we can’t be sure whether poor mental health was the cause of 'bad behaviour' and thus the reprimand or suspension, rather than the consequence.

Figure 1. Risk of bias in included studies in the review.

Effect of disciplinary strategies on pupil mental health and wellbeing outcomes

•   Depression

i.   School Suspension Strategies

Five studies^{29,32,33,35,38} reported on depression due to suspension.

Odds of greater total difficulties (SDQ) were found to be significantly higher for those who were suspended in two studies^32,35. Rushton et al.³⁸ also found that being suspended from school was associated with increased odds of depressive symptomology. Two studies^29,33 found self-reported depressive symptoms were not related to suspension.

ii.   Verbal reprimand strategies

None of the two studies assessing verbal punishment strategies reported on depression.

iii.   Mixed strategies

Two studies^28,30 reported depression assessing two slightly different strategies that included suspensions and some other forms of punishments together. Both found that punishment-based policies led to more depressive symptoms, but at different time points.

Chen et al.²⁸ found that Black pupils attending a school that disproportionately punished Black students had greater depressive symptoms as an adult ten years later (β= 0.11,95% CI: 0.04, 0.18), compared to their White peers. Eyllon et al.³⁰ found strict (vs lenient) policies to increase depression in pupils so that each unit increase in school’s policy being strict led to a 1.03 unit rise in pupil depression scores on average (95% CI: 0.15, 1.91).

Impact on population subgroups

The two studies reporting data on ethnicity were not in agreement. While Chen et al.²⁸ found Black pupils to be disproportionately affected by disciplinary punishments and the consequent depression in later age, Eyllon et al.³⁰ found no link between ethnicity and higher depression due to strict school policies within one year.

•   Anxiety

No suspension or mixed strategy studies reported this outcome.

A single study set in Poland³⁴ found that higher school stress brought on by verbal reprimand strategies led to higher anxiety in pupils (R = 0.30, p <.001).

•   Psychiatric disorder

A single study³² found children who had been suspended from school had higher odds of diagnosis of a new psychiatric disorder (OR 7.09; 95%CI 5.07 to 9.91; p < 0.001) compared to those not suspended.

No studies of verbal reprimand or mixed strategies reported this outcome.

•   General mental wellbeing

One study³¹ assessing the effect of suspension found a non-significant (p=0.15) lower wellbeing (on Warwick Edinburgh Mental Wellbeing Scale) and significantly (p=0.003) greater use of mental health services for suspended children compared to pupils who have never been suspended. Another found children who were suspended were twice as likely to have poor mental health³⁷.

A single study³⁶ assessing link between verbal disciplinary strategies and pupil mental wellbeing (on author developed scale) reported no outcome data.

No study on mixed strategies reported this outcome.

•   Internalising symptoms

One study on suspension³⁹ found that each additional suspension per school led to increase in internalising scores by 0.05%, while another²⁹ found that suspensions were not associated with internalising problems.

No verbal reprimand or mixed strategy studies reported this outcome.

•   Externalising symptoms

A single study found that in schools which suspended proportionally more Black pupils than White pupils, Black pupils overall showed higher externalising symptoms²⁷.

No verbal reprimand or mixed strategy studies reported this outcome.

Effect of disciplinary strategies on pupil social and behavioural outcomes

i.   School Suspension Strategies

Three studies^29,35,40 found suspension was associated with poorer (more disruptive, less pro-social) behaviour.

Two studies found suspensions were associated with lower perception of social belonging at school^27,35.

School-level Black–White suspension gaps (i.e., excess risk of out-of-school suspension among Black students relative to White students,) were associated with Black students’ perceptions of less school equity in a single study²⁷.

A single study³³ found suspensions led to greater marijuana use but had no association with tobacco or alcohol use. A single study³⁹ found no association between out-of-school suspension and self-esteem.

ii.   Verbal reprimand strategies

A single study (Roache) found that aggressive verbal punishments from teachers led to increased disruptive behaviour in the classroom (r=0.48, p<0.05), being more distracted from class work (r=0.72, p<0.05), and reduced pupil interest in the subject being taught (r=-0.58, p<0.05).

No studies in this category reported social outcomes.

iii.   Mixed strategies

      No studies of mixed strategies reported social outcomes.

Effect of disciplinary strategies on academic outcomes

Of the three studies assessing educational outcomes, one comprehensively reported data and³³ found no link between suspension and grade scores. This effect remained non-significant (although direction was opposite) after adjusting for demographic factors including ethnicity. One study³⁷ did not report data on the effect of suspension on grades, and the other⁴⁰ said they found lower scores on a composite of academic performance habits and skills but did not report data to support this finding.

Piekarska et al. 34 found that verbal punishments from teachers caused school stress which negatively impacted academic performance as grade point average.

While Chen et al. (2021) reported no direct effect of greater punishment on long-term educational attainment, they did find that for children who were not academically oriented, greater punishment was associated with lower educational attainment. This study included only Black American pupils.

Summary of findings

Our review illustrates that evidence on the impact of disciplinary strategies in schools is scarce and of low quality. Although at high risk of bias, five out of seven studies assessing depressive symptomatology found it to be associated with exposure to disciplinary strategies. All three studies on general mental wellbeing found it to be associated with exposure to disciplinary strategies. Single studies on anxiety, psychiatric disorder diagnosis, and externalising symptoms also found that disciplinary strategies were associated with these issues. Internalising symptoms, and a similar link with externalising symptoms, were only seen in one of the two studies to be associated with a disciplinary approach. Similar effect was seen with social outcomes where, overall, disciplinary strategies were associated with poor social behaviour (n=4), lower school belonging (n=2), and greater marijuana use (n=1), but had no association with tobacco use or self-esteem (n=1). Evidence of the impact on educational attainment was limited and it was not clear how they were related to disciplinary strategies.

Limitations of our review

We followed PRISMA standards when reporting the review and searched comprehensively using relevant scientific databases and grey literature sources. We were inclusive in our criteria for studies to allow us to examine the full range of effects of these commonly used strategies. Considering how widespread their use is, the empirical evidence on these strategies is limited for wellbeing, behaviour or academic outcomes.

Included studies were at high risk of bias in most domains. This is a major limitation of our findings. There is a need for better quality research to address these questions.

We searched for, and included studies that reported at least one primary (mental wellbeing) outcome. Our restriction to primary outcome reporting has likely overlooked evidence on educational and behavioural outcomes reported in studies without a focus on general mental wellbeing. Thus, our findings on these outcomes are likely not to be comprehensive, although they may be indicative of the general trend. This review included studies conducted only in mainstream schools and therefore the findings do not extend to other settings. However, when screening the literature, there were studies that focused on specialist schools or exclusively including pupils with additional learning needs (e.g., attention deficit hyperactivity disorder) and these should be assessed in a separate review.

We included all author definitions for strategies (e.g., suspension or temporary exclusion could be anything from a few hours to several days, and may or may not include supervised confinement to a room or location in school) to not miss any relevant evidence. There is however a lack of clear definitions and descriptions for any of the disciplinary strategies. Thus, there is a need to clearly define these interventions and their proposed impact before research on these can give clear conclusion on their relative impacts. For example, UK defines suspensions as any fixed time exclusion between one school period (length of a lesson varies from 30 minutes to 120 minutes) and 45 school days⁴⁶. This definition may be different from those used in other nations. This would invariably also be reflected in studies from different countries. We would anticipate that the effect of a 45-minute isolation may be different from that of a week-long or month-long suspension. A differential or dose response effect may only be elicited if the definitions used in each study are clear.