Discussion
The present evidence review has examined the effectiveness of fire safety interventions aimed at promoting fire safety knowledge, skills, attitudes and/or behaviours in preschool children. The literature in this area is characterised by a range of research designs, including uncontrolled post-assessment studies, single-subject studies, controlled studies without pre-testing, an uncontrolled before-and-after study, and randomised controlled studies. Thus, the evidence reflects a high degree of variation in methodological robustness. A key limitation of the uncontrolled post-assessment studies is a reliance on qualitative outcome assessment methods, which are highly susceptible to response biases (such as participants responding in ways they think that the researchers want or expect, emotions, judgements, or perceptions, and memory errors) and biases in data interpretation. Qualitative assessments are also typically associated with very small sample sizes. Small samples (especially fewer than 100 cases) reduce the likelihood of the data distribution of the sample matching that of the wider population. Furthermore, numerous studies did not include a control group. Accordingly, in those studies, it is not possible to infer causal links between the intervention and the outcomes, which could have been caused by extraneous variables such as maturation, concurrent educational initiatives or random events. In addition, a number of evaluations assessed outcomes after the intervention but not before. Therefore, it is impossible to appraise whether the intervention was linked to any change in the outcomes, and significant differences yielded between intervention and control groups at post-assessment could potentially be explained by baseline differences. For the purpose of making causal inferences, the strongest research design used was the RCT. However, the RCT studies also need to be interpreted with caution. In some of these studies, a relatively small number of clusters were randomly allocated to intervention or control conditions, so group differences in the outcomes could potentially be explained by extraneous differences amongst the clusters rather than the intervention itself. In addition, in one RCT (Mori and Peterson, 1986), all participants received a version of a fire safety intervention, thus the effect of the intervention as a whole was not assessed.
The RCTs reviewed constitute strong evidence for the claim that it is possible to improve preschoolers’ fire safety-related knowledge, behavioural intentions, behavioural skills and behaviours through interventions, despite their developmentally limited neuropsychological abilities. In addition, the less robust studies are consistent with this claim. The weaker studies also provide tentative evidence that it might be possible to shape fire safety attitudes; however, future research will need to adopt an RCT design to explore effects. Furthermore, the evidence suggests that the fire safety behavioural skills preschoolers acquire may generalise across contexts. In particular, it suggests they may generalise: across settings, such as from school to the home environment; and from artificial to naturalistic contexts, such as from the context of a researcher-induced fire alarm sound to a school-wide fire drill, and from a virtual video game world to the real world (Coles et al., 2007; Morgan, 2017; Padgett, Strickland and Coles, 2006; Rossi et al., 2017; Vanselow and Hanley, 2014). However, most of these studies were conducted with atypically developing preschoolers, so the findings in those studies may not be applicable to typically developing preschoolers. In addition, most of these studies lacked a control group. One was an RCT (Coles et al., 2007) but did not pre-assess fire safety skills in the generalised context or assess these skills amongst the controls. Accordingly, further RCTs are needed to examine causal links between fire safety interventions and generalised safety skills in typically developing preschoolers. None of the identified studies assessed whether enhanced knowledge, skills, attitudes and intentions translated into appropriate behavioural responses in the context of a real fire emergency. However, confidence that this translation could occur is increased by behaviour change theory applicable to young children (e.g., Cane, O’Connor and Michie, 2012; Cibrian, Tentori and Martínez-García, 2016), and the indicative evidence of skill generalisability found in the present study. However, future research will need to subject this hypothesis to empirical assessment.
Whilst there was variation across the studies in the extent to which the interventions had met their respective aims, it is difficult to appraise the relative effectiveness of the different interventions due to a high level of study variability. Nonetheless, despite the variation in methodological strength, all of the studies were at least consistent with the claim that fire safety interventions lead to relevant positive outcomes. Each RCT that included a control group that did not receive a fire safety intervention found salutary impacts of the intervention as a whole. Those RCTs included individual interventions with preschoolers, and group interventions with preschoolers in the school or child care setting. Individual interventions in those RCT studies used video games to convey information. The video games avoided placing reading demands on the child and were able to be completed by the child independently. Group interventions in those RCT studies used simulated telephone apparatus. Furthermore, individual and group interventions in those RCT studies used storybooks and illustrations, including colourful, cartoon, two-dimensional or three-dimensional imagery and photographs. Previous research indicates that illustrations are associated with enhanced comprehension, attention and memory for material amongst young children (Evans and Saint-Aubin, 2005; Greenhoot and Semb, 2007; Walsh, 2003). Owing to a lack of systematic manipulation, it is not possible to draw definitive conclusions about causal links between the intervention elements mentioned and the relevant salutary intervention impacts found; however, their association with positive effects is consistent with causality. In the weaker studies, individual interventions also used video, pictorial stimuli and ostensibly dangerous physical objects, and group interventions also used video, toys, audio materials, posters, visual cue cards and educational websites. Future RCTs will need to establish whether these intervention elements are connected to salutary impacts.
In one intervention, which was evaluated in an uncontrolled before-and-after study, the classroom environment was shaped such that physical learning aids were available to the children and learning opportunities were embedded (Jones, 2006). This approach was based on the Vygotskian theory that learning is entwined with the social and psychical environment (Vygotsky, 1993). Although causal conclusions are precluded due to the study design, greater accessibility of the materials was linked to more reflective dialogue (discussed further below), and the intervention as a whole was associated with improved fire safety knowledge (Jones, 2006). In addition, the presence of the fire service was a strategy used by group interventions in some of the less robust studies. However, none of the identified studies examined through systematic manipulation whether fire service presence had positive effects. Given that fire service presence was not deployed in any of the RCT studies, this strategy does not appear to be essential for salutary impacts.
Furthermore, parental involvement was incorporated by individual and group interventions in some of the RCTs and weaker studies. In the RCTs, parental involvement took the form of support with storybook reading and engagement in home-based activities, such as devising a home escape plan. One study systematically manipulated parental involvement, and found that parental mediation regarding intervention video material was associated with superior outcomes (Borzekowski et al., 2014). However, due to a lack of pre-testing, causal inferences cannot be made. Previous research has demonstrated the benefits of parental involvement in early years education (Reynolds and Shlafer, 2010). In the preschool years, children spend more time with parents and relatives than with other social actors; thus, parental behaviour is a potent mechanism through which to shape child development (Reynolds and Shlafer, 2010). Furthermore, given that parents accompany their children in numerous different contexts, parental involvement is likely to facilitate the generalisation of skills across multiple domains, including the home environment, which is a particularly important context from a fire safety perspective. Although the reviewed evidence provides indicative support for positive effects of parental involvement in fire safety interventions, further RCTs that systematically manipulate this element are needed.
Some of the reviewed evidence indicates that, across intervention types, providing clear guidance to those involved in the implementation of interventions, including teachers as well as parents, is linked to superior outcomes. One RCT systematically manipulated guidance. It found that the positive effects of a teacher-delivered emergency dialling intervention were significantly greater where teachers had been provided with specific instructions in how to deliver the training, compared with where training had been devised by the teachers (Jones and Kazdin, 1980). Consistent with this, in a controlled study without a pre-test, the findings of superior outcomes associated with parental mediation regarding video material were most robust where parental mediation had been scripted by the researchers (Borzekowski et al., 2014). However, these findings conflict with those of another RCT which manipulated guidance (Mori and Peterson, 1986). In that study, which evaluated a teacher-delivered intervention on safety response rules, improvement in preschoolers’ fire safety knowledge did not differ significantly between a condition in which the teacher had received a training manual plus professional supervision and a condition in which they had received the training manual only. Although interpretation of the evidence is difficult due to heterogeneity across studies, it is possible that the training manual in Mori and Peterson’s (1986) study was sufficient, hence the lack of significant differences across conditions.
Repeated exposure to intervention material over time also appears to be connected to better outcomes. Lidstone (2006) found that, after a one-off 45 minute puppet performance in a school assembly, the children were able to recite key safety message slogans but their understanding of the meaning of those messages was relatively low. On the other hand, Reim (2011) found that greater exposure to storybook intervention material was significantly associated with greater improvements in knowledge of home hazards. Similarly, Morrongiello et al. (2012) found that more time playing a fire safety video game was marginally significantly associated with a greater degree of improvement in fire safety knowledge and behavioural intentions. However, the validity of Lidstone’s (2006) evaluation is reduced by the qualitative methods that were used, and Reim’s (2011) and Morrongiello et al.’s (2012) findings are correlational and therefore do not support causal inferences. Nevertheless, the findings align with previous research in different contexts, which has found that repeated exposure to material is associated with enhanced comprehension amongst preschoolers (e.g., Crawley et al., 1999). Potential means of promoting repeated exposure include delivering messages via multiple mediums, including mediums that children enjoy using and will engage with repeatedly (e.g., video games), and implementing learning programmes that convey information recurrently over the child’s developmental course (Lidstone, 2006; Morrongiello et al., 2012). Future research should systematically examine the effects of repeated exposure to fire safety interventions.
Moreover, individual and group interventions in some of the RCTs and weaker studies incorporated experiential and/or interactive activities, such as: interpersonal interaction including one-to-one training, question-and-answer sessions, dramatic play and interactive storybook reading; virtual gameplay; real-world games; behavioural rehearsal; exploration of physical objects; singing; and crafts. Due to the presence of confounding variables, it is not possible to conclude that these activities were causally linked to positive outcomes. However, it is notable that, in one controlled study without a pre-test, the recipients of an intervention, which lacked experiential activities and in which children passively received aural-visual information, demonstrated significantly better knowledge of only a limited range of fire issues, and did not differ significantly in fire safety attitudes, compared with controls (Folkman and Taylor, 1972). The notion that activity-based interventions have positive effects is consistent with experiential learning theory (Kolb, 2014; Kolb and Fry, 1974), which posits that, through activities, children engage in a cyclical process of action and reflection which enables learning. Additionally, presenting material to children in an interactive manner has been found to facilitate the assimilation of new information, thereby enhancing cognitive outcomes (Chi et al., 1994; Morrongiello, Miron and Reutz, 1998). Interaction enables the delivery of feedback, which was associated with positive outcomes in the reviewed evidence. Overall, experiential and interactive activities in fire safety interventions with preschoolers are likely to have positive effects. Although the reviewed evidence is consistent with this, methodologically robust research that systematically manipulates intervention activities is required to strengthen conclusions.
A number of the individual and group interventions, evaluated by RCTs and weaker studies, deployed BST or variants thereof in order to impart behavioural skills and thereby enable positive fire safety behaviours. BST involves instruction, modelling, rehearsal and feedback (Rossi et al., 2017). Due to a lack of systematic manipulation, it is not possible to conclude from the RCTs which elements of the BST, if any, contributed to the positive intervention impacts. Nonetheless, the association of BST with salutary behavioural outcomes aligns with notional positive effects. Previous research has found positive effects of BST with children (Miltenberger, 2008). Furthermore, as mentioned earlier, some single-subject studies and an RCT that lacked pre-assessment, which were predominantly conducted with atypically developing preschoolers, yielded some indicative evidence that it may be possible for safety behavioural skills acquired through BST to generalise across contexts amongst preschoolers (Coles et al., 2007; Morgan, 2017; Padgett, Strickland and Coles, 2006; Rossi et al., 2017; Vanselow and Hanley, 2014). However, previous research indicates that the effects of BST with children do not always generalise to different contexts (Miltenberger, 2012). Consonant with this, one single-subject study found that, whilst some preschoolers achieved competence in responding safely to ostensibly dangerous physical stimuli after playing a video game featuring BST, some preschoolers only reached competence after additional in-situ training, whereby children received BST in the real world (Vanselow and Hanley, 2014). This study provides tentative evidence suggesting that the supplementary use of in-situ training may, by facilitating skill generalisation, augment improvements in safety behaviours engendered by non-naturalistic BST. However, controlled and systematic research is needed to reach conclusions about the effectiveness of different BST techniques.
Additional intervention techniques employed for intervention purposes included forms of scaffolding (Coles et al., 2007; Jones, 2006; Mori and Peterson, 1986), which was associated with positive intervention impacts in an RCT (Coles et al., 2007). Scaffolding involves reducing the level of support provided to the child as they attain more competence. It is grounded on the Vygotskian perspective that children possess capabilities that are in the process of maturation, and that these develop through their task-directed implementation with the appropriate level of assistance (Vygotsky, 1978). In addition, encouragement of reflective dialogue was a technique used in one intervention with preschoolers (Jones, 2006). Reflective dialogue denotes the ability to self-reflect on thinking and learning, and coordinate decision-making and action on the basis of that process (Cullen, 1995). Amongst preschoolers, it can manifest as conversation regarding acquired knowledge. Jones’ (2006) uncontrolled before-and-after evaluation found that fire safety knowledge improved, which is consistent with – though does not constitute direct evidential support for – positive effects of reflective dialogue promotion. Similarly, previous research has identified that reflective dialogue promotion is connected with better learning outcomes (Cullen, 1995). An inductive reasoning approach to teaching was a further intervention technique used in one intervention (Morrongiello et al., 2016). This was characterised by an emphasis on cause-and-effect relationships between unsafe behaviours and injuries. The empirically supported basis for this approach is that younger children are more likely to reason inductively rather than deductively, reaching inferences based on specific experiences (Galotti, Komatsu and Voelz, 1997). The intervention was found to have positive impacts on fire safety knowledge in the RCT (Morrongiello et al., 2016), which is consistent with notional positive effects of this technique. Previous research has found positive outcomes associated with the inductive reasoning approach (e.g., Coppens, 1986). Furthermore, one controlled study without a pre-test found that focusing on the positive outcomes of correct behaviours (gain-framing) was associated with superior outcomes at post-assessment compared with a focus on the negative outcomes of incorrect behaviours (loss-framing; Borzekowski et al., 2014). Future research should examine the potential salutary influence of gain-framing using an RCT design. Finally, based on the Health Belief Model (e.g., Gielen and Sleet, 2003), some interventions, evaluated by RCTs which found positive intervention impacts, conveyed messages which sought to promote perceptions of personal vulnerability, perceptions of threat, and/or self-efficacy regarding fire safety capabilities (Morrongiello et al., 2016; Reim, 2011). Future research should examine the effects of these messages in fire safety interventions through systematic manipulation.
In addition to those already mentioned, a number of limitations should be borne in mind when interpreting the results. Firstly, where applicable, follow-up periods tended to be relatively short; thus, it is not possible to reach conclusions about the long-term maintenance of positive effects. Secondly, the majority of the studies originated from the USA, which restricts the generalisability of the results. Thirdly, as is often the case in psychological research, studies tended to use convenience samples, which increases the risk that the samples were not representative of the broader population, further restricting generalisability.