Knowing is not enough: an in situ evaluation of personal safety behaviour in individuals with intellectual disabilities

According to The State of the World’s Children (UNICEF, 2021), more than 1.1 million children are reported missing or abducted globally each year. Within the European Union alone, approximately 250,000 children go missing annually, equating to one child every two minutes (Missing Children Europe, 2020). These figures highlight the urgent need for effective prevention strategies, particularly for vulnerable populations such as children, women and individuals with disabilities (Farid et al., 2024).

Contrary to common public perceptions, most child abductions do not involve overt violence. Instead, perpetrators often establish a brief, seemingly friendly interaction and use lures or incentives to gain compliance (Miltenberger and Olsen, 1996). This method is particularly effective among individuals who may have difficulty interpreting social cues or assessing risk.

Research consistently demonstrates that individuals with intellectual disabilities are at significantly higher risk of abuse and victimisation compared to their typically developing peers (Fisher et al., 2016; Horner-Johnson and Drum, 2006; Lee et al., 2008; Lund and Vaughn-Jensen, 2012; Wilson and Brewer, 1992). For adolescents with disabilities, this risk increases substantially, with some studies suggesting rates up to four times higher than those of the general population (Simmons et al., 2023).

Despite the well-documented vulnerability of this population, relatively few studies have focused on personal safety skills, particularly in relation to responding to stranger lures (Akmanoglu and Tekin-Iftar, 2011; Godish et al., 2017; Trevor et al., 2021). A review of empirical studies identified only a small number of interventions specifically targeting responses to stranger approaches (Collins et al., 1992; Gast et al., 1993; Haseltine and Miltenberger, 1990; Mazzucchelli, 2001; Watson et al., 1992).

Importantly, existing research suggests that knowledge of safety rules does not necessarily translate into appropriate behaviour in real-life situations. Studies with typically developing children have shown that a large proportion would follow a stranger when presented with an attractive lure (Poche et al., 1981; Poche et al., 1988; Marchand-Martella and Huber, 1996; Sanchez and Miltenberger, 2015; Author 3 & Author 2, 2023).

The increased vulnerability of individuals with intellectual disabilities has been attributed to multiple factors, including dependence on others, deficits in communication and social skills and difficulties in recognising others’ intentions (Algood et al., 2011; Foster and Sandel, 2010; Riby et al., 2014). At an individual level, factors such as limited decision-making abilities, poor understanding of interpersonal boundaries and increased compliance further heighten risk (Mahoney and Poling, 2011; Martinello, 2015; McEachern, 2012).

Given these challenges, the development and evaluation of effective personal safety interventions is critical. However, an important gap remains between assessing what individuals know and what they actually do in real-life situations. The present study therefore addressed the following research question: To what extent do individuals with mild to moderate intellectual disabilities demonstrate appropriate personal safety behaviours in response to stranger lures in real-life conditions, and how does their in situ behavioural performance compare to their knowledge of personal safety skills as assessed through a structured interview tool?

The study involved 89 individuals with mild to moderate intellectual disabilities (61 males, 28 females; mean age = 24.5 ± 5.2 years), recruited through convenience sampling.

Inclusion criteria were: (a) an official diagnosis of mild or moderate intellectual disability, (b) enrolment in one of the participating social welfare centres and (c) the ability to understand simple verbal instructions. Individuals with a dual diagnosis (e.g. intellectual disability co-occurring with autism or ADHD) were not excluded from participation, provided they met the above criteria. Exclusion criteria were: (a) a diagnosis of severe or profound intellectual disability, (b) inability to understand simple verbal instructions as judged by centre staff and (c) absence of written consent from a parent or legal guardian.

Participants were identified as having intellectual disabilities based on official disability certification issued by the national disability assessment authority. These certifications included diagnostic classification and level of intellectual disability (mild or moderate), determined through standardised clinical and psychological assessment procedures conducted by certified multidisciplinary evaluation committees.

For the purposes of this study, classification of intellectual disability was based on the diagnostic information reported in these official certificates, which are required for enrolment in the participating social welfare centres.

Participants were recruited from two social welfare centres located in major cities in Northern Greece. These centres provide structured daily programmes focusing on life skills training, social integration and vocational preparation. Participants attend on a daily basis and engage in group-based educational and recreational activities supervised by special education teachers, occupational therapists, psychologists and support staff. Both centres follow a person-centred approach to support, emphasising functional skill development, community participation and quality of life. Neither centre used a specific manualised therapeutic model (e.g. applied behaviour analysis or positive behaviour support) as a primary framework at the time of the study; rather, support was delivered through individualised programmes developed collaboratively by multidisciplinary teams in line with national legislative requirements for social welfare services in Greece.

All assessment procedures were conducted within the familiar environment of these centres to enhance ecological validity and reduce potential anxiety associated with unfamiliar settings.

The present study used a cross-sectional observational design to evaluate the relationship between knowledge of personal safety skills and actual behavioural responses in real-life conditions.

Participants’ knowledge of personal safety skills was assessed using the What-If Situation Test (WIST III-R).

The WIST is a structured interview-based assessment tool designed to evaluate individuals’ responses to hypothetical risky situations. It assesses three core behavioural components:

1. SAY: verbal refusal (score range: 0–2).

2. DO: physical avoidance (score range: 0–2).

3. TELL: reporting the incident to a trusted person (score range: 0–4, plus 1 additional point for persistence in reporting).

An additional item assessed whether the participant recognised the situation as unsafe (scored 0–1).

Total scores ranged from 0 to 10, with higher scores indicating greater knowledge of appropriate safety responses. Scores were interpreted as follows (Yu et al., 2017):

• 0: extremely low knowledge.

• 1–5: below average.

• 6–9: above average.

• 10: excellent.

Two scenarios were used:

1. A stranger offering a reward after receiving help.

2. A stranger inviting the participant to follow them to a car with a desirable item.

The WIST has demonstrated adequate reliability (0.70–0.79) and internal consistency (0.80–0.89) (Jin et al., 2016; Yu et al., 2017).

To evaluate real-life behavioural responses, an in situ assessment was conducted using a structured and standardised protocol.

This method is widely considered the most valid approach for assessing whether individuals can apply safety skills in real-world conditions (Gatheridge et al., 2004; Himle et al., 2004).

Each participant was accompanied by a staff member to a familiar outdoor setting (e.g. park, courtyard). The staff member then briefly left the participant alone under the pretext of completing a task, instructing them to remain in place.

Shortly thereafter, a trained confederate (female adult unknown to the participant) approached and initiated interaction using one of several scripted scenarios, including:

• requesting help locating a lost item;

• offering a reward; and

• claiming to have been sent by a trusted authority figure.

The confederate was trained to follow a standardised script to ensure consistency across participants.

During the interaction, the confederate:

• attempted to obtain personal information;

• introduced a lure (e.g. toy, sweets, reward); and

• encouraged the participant to follow them (e.g. to a car).

If the participant agreed to follow, the interaction was terminated before leaving the safe area.

Participants’ responses were scored as follows:

• 0: agreed to follow the stranger.

• 1: refused to follow but did not verbally refuse.

• 2: verbally refused but did not leave or report.

• 3: refused and left but did not report.

• 4: refused, left and reported the incident.

Participants scoring:

• 3–4: considered to have acquired the safety skill.

• 2: emerging skill.

• 0–1: no skill acquisition.

Ethical approval for the study was obtained from the institutional ethics committee, and all procedures complied with the principles of the Declaration of Helsinki.

Written informed consent was obtained from parents or legal guardians, as all participants were under protective supervision.

Given the use of in situ assessment involving elements of deception, several safeguards were implemented:

• All scenarios were carefully designed to minimise psychological distress.

• Participants were not exposed to actual danger at any point.

• The interaction was terminated before participants left the supervised environment.

• A staff member remained nearby to ensure safety.

• Participants were debriefed immediately after the procedure, and the purpose of the scenario was explained in accessible language. Given that prior disclosure of the study’s true purpose would have invalidated the in situ assessment, formal assent from participants themselves could not be obtained beforehand. To address this, post hoc assent was sought from each participant following debriefing. Support staff assisted in explaining, in accessible language, that participation had formed part of a research study and that each individual had the right to withdraw their data if they did not wish their responses to be included. No participant exercised this right.

To ensure confidentiality and data protection, each participant was assigned a unique code. Identifiable information was accessible only to the research team and used exclusively for research purposes.

Descriptive statistics (frequencies, percentages, means and standard deviations) were used to summarise participants’ performance on the WIST and in situ assessment.

To examine associations between categorical variables (e.g. gender and behavioural response), chi-square (χ²) tests of independence were conducted.

In addition, the relationship between knowledge (WIST scores) and behavioural performance (in situ scores) was explored using correlation analysis. Prior to analysis, the normality of score distributions was assessed using the Shapiro–Wilk test. Results indicated that in situ behavioural scores were not normally distributed (W = 0.71, p < 0.001), reflecting the highly skewed nature of responses. Accordingly, Spearman’s rank correlation coefficient was used to examine the relationship between WIST scores and in situ performance.

Statistical significance was set at p < 0.05.

Descriptive statistics indicated that the majority of participants demonstrated relatively high levels of knowledge regarding personal safety skills.

Specifically, the majority of participants demonstrated relatively high levels of knowledge, with 94% scoring within the above-average range (6–9), 4.5% scoring below average (1–5) and only 1% achieving an excellent score (10), while no participant scored in the extremely low range (0) (see Table 1). These findings suggest that most participants were able to verbally identify appropriate safety responses in hypothetical situations.

The in situ assessment revealed a markedly different pattern compared to the WIST results. Out of 89 participants, 80 (89.88%) agreed to follow the stranger, while only 9 (10.11%) refused (Table 2). This indicates that the overwhelming majority of participants failed to apply safety skills in a real-life context.

As shown in Table 3, compliance rates remained high across all conditions, indicating that no type of lure significantly reduced vulnerability. Across all conditions, compliance rates remained extremely high, indicating that no type of lure significantly reduced vulnerability.

Although males appeared slightly more likely to follow, the difference was not statistically significant: χ²(1, N = 89) = 1.34, p > 0.05, Cramér’s V = 0.12 (small effect size). These findings suggest that vulnerability to stranger lures was high across both genders (Table 4).

Participants’ performance according to the behavioural scoring system revealed very low levels of skill acquisition. Notably, no participant demonstrated a fully appropriate response (score = 4), and only three participants (3.7%) approached mastery level (score ≥ 3) (Table 5).

Despite the high levels of knowledge observed in the WIST assessment, behavioural performance in the in situ assessment was extremely poor.

A correlation analysis between WIST scores and in situ performance indicated no meaningful association between knowledge and actual behaviour (non-significant relationship).

This finding suggests that possessing knowledge of safety rules does not necessarily translate into the ability to apply these skills in real-life situations.

The present study aimed to evaluate both the knowledge and the actual behavioural performance of individuals with intellectual disabilities in response to potentially dangerous situations involving stranger lures.

The findings revealed a striking discrepancy between knowledge and behaviour. Although the majority of participants demonstrated above-average knowledge of personal safety skills, nearly nine out of ten participants followed a stranger during the in situ assessment. Notably, no participant demonstrated a fully appropriate safety response involving refusal, avoidance and reporting.

These findings are consistent with previous research indicating that knowledge of safety rules does not necessarily translate into effective real-world behaviour (Author 3 & Author 2, 2023). The results extend this literature by demonstrating that this gap is particularly pronounced among individuals with intellectual disabilities.

The observed discrepancy between knowledge and behaviour can be understood through a performance deficit framework, which distinguishes between knowing what to do and being able to do it under real-life conditions. In real-world situations, individuals are required to process multiple dynamic and often unpredictable cues, which are typically absent in hypothetical scenarios. For individuals with intellectual disabilities, this increased contextual complexity may result in cognitive overload, thereby impairing decision-making under pressure. In addition, heightened levels of social compliance and a tendency to trust others − particularly when approached in a friendly and non-threatening manner − may further contribute to unsafe responses, overriding previously learned safety rules (Dunne and Power, 1990; Goldfarb et al., 2008; Sanchez and Miltenberger, 2015).

Another critical factor relates to difficulties in the generalisation of learned skills. It is well established that individuals with intellectual disabilities may struggle to transfer knowledge acquired in structured or instructional settings to novel, real-life contexts. As a result, responses that are correctly identified in assessment tools such as the WIST may not be effectively enacted in practice. Finally, misconceptions about the nature of “strangers” may also play a role. Individuals may rely on stereotypical representations of danger, expecting threatening individuals to appear visibly suspicious. In the present study, the use of a friendly female confederate may have reduced perceived risk and contributed to the high levels of compliance observed. Research indicates that both children and adults tend to associate danger with specific physical features − expecting threatening individuals to appear visibly aggressive or unkempt − rather than evaluating behaviour (Yarmey, 1993; Todorov et al., 2009; Cogsdill and Banaji, 2015). Moreover, female faces displaying positive affect are consistently rated as more trustworthy than male, unattractive or angry faces (Ma et al., 2015; Todorov et al., 2009), which may explain why a friendly female confederate did not trigger avoidance responses among participants.

The findings of the present study have important implications for practitioners, educators and policymakers working with individuals with intellectual disabilities. The results clearly indicate that traditional approaches to safety education, which primarily focus on teaching rules such as “do not talk to strangers”, are insufficient for promoting safe behaviour in real-life contexts. Instead, there is a need to shift towards interventions that emphasise behavioural performance and the generalisation of skills across settings.

One of the most effective approaches for achieving this is behavioural skills training (Miltenberger, 2008), which incorporates instruction, modelling, rehearsal and feedback. Crucially, training should include in situ practice, allowing individuals to apply learned skills in realistic environments, thereby increasing the likelihood of generalisation. Interventions should prioritise the development of functional safety skills, particularly the ability to refuse inappropriate requests, move away from potentially dangerous situations and report incidents to a trusted person. These skills must be practised repeatedly across different contexts to ensure maintenance and transfer.

Furthermore, educational programmes should move beyond simplistic notions of “stranger danger” and instead focus on helping individuals recognise unsafe behaviours and understand common manipulation strategies, such as the use of lures or requests for secrecy. Finally, the protection of individuals with intellectual disabilities should be viewed as a shared responsibility. Effective safeguarding requires coordinated efforts among families, educators, healthcare professionals and policymakers, highlighting the need for a multi-agency approach.

These findings suggest that traditional didactic approaches may be insufficient to promote behavioural change. There is a clear need for more experiential and embodied forms of learning, such as drama-based interventions and immersive training approaches, which allow individuals to actively engage with realistic scenarios. Such approaches may enhance skill generalisation and bridge the gap between knowledge and behaviour.

The extremely high rate of compliance observed in this study raises serious concerns regarding the safety, autonomy and overall well-being of individuals with intellectual disabilities. These findings highlight the urgent need for the systematic implementation of structured safety training programmes within educational and social care settings.

In particular, personal safety skills should be formally integrated into life-skills curricula and recognised as a fundamental component of independent living. At the policy level, there is a clear need for the development of evidence-based national guidelines that address the prevention of victimisation in this vulnerable population. Enhancing personal safety skills is directly associated with increased independence, improved quality of life and reduced risk of abuse, making it a critical priority for both practice and policy.

Several limitations should be considered when interpreting the findings of this study. Firstly, the use of convenience sampling limits the generalisability of the results to the broader population of individuals with intellectual disabilities. Secondly, the study was conducted within a specific cultural context (Greece), and cultural norms may influence both social behaviour and perceptions of strangers.

Although the use of in situ assessment enhances ecological validity, it nevertheless represents a structured simulation and may not fully capture the complexity of real-life risk situations. In addition, the use of a single confederate profile (female adult) may have influenced participants’ responses, as different outcomes might emerge with alternative profiles, such as male individuals or authority figures. Finally, the study focuses exclusively on baseline assessment and does not examine long-term skill retention or the effects of intervention, which are addressed in subsequent phases of the broader research project. Furthermore, the use of a post hoc assent procedure, while ethically necessary given the deceptive nature of the in situ assessment, represents a limitation in terms of fully informed participation. The extent to which individuals with intellectual disabilities were able to meaningfully exercise their right to withdraw data following debriefing warrants further consideration. In addition, participants were recruited exclusively from two social welfare centres in Northern Greece; this contextual specificity may limit the generalisability of findings to individuals attending other types of services or residing in different cultural or geographic settings.

Future research should build upon the findings of the present study by examining the effectiveness of intervention programmes designed to improve real-life safety behaviours among individuals with intellectual disabilities. In particular, further investigation is needed into the extent to which safety skills generalise across different settings, contexts and types of social interactions.

Additional research should explore potential differences across levels of intellectual disability, as well as include younger age groups, given the importance of early intervention in the development of safety skills. It would also be valuable to assess responses to a wider range of individuals, including authority figures such as police officers or healthcare professionals, to better understand situational variability in behaviour.

Moreover, future studies should expand the scope of safety skills to include other critical domains, such as road safety, emergency response and digital safety. Finally, cross-cultural research would provide important insights into how social norms and environmental factors influence vulnerability and behavioural responses, thereby contributing to the development of more effective and culturally sensitive intervention strategies. Importantly, future research should also address the teaching of more nuanced personal safety skills that reflect the full range of risks faced by individuals with intellectual disabilities. The evidence consistently indicates that abuse is more commonly perpetrated by individuals known to the person than by strangers. Future interventions should therefore extend beyond stranger-lure scenarios to include skills such as recognising inappropriate behaviour by familiar individuals, boundary-setting, help-seeking and responding to risks within trusted relationships. Abduction by strangers, while a legitimate concern, is comparatively rare, and typically arises only after an individual has first become separated or lost; in such cases, preventative skills such as remaining close to a caregiver or support person may be more protective.

The present study provides compelling evidence of a substantial gap between knowledge and behavioural performance in personal safety situations among individuals with intellectual disabilities.

Although most participants demonstrated adequate knowledge of appropriate safety responses, the vast majority failed to apply these skills when confronted with a realistic scenario involving a stranger lure. This discrepancy highlights a critical limitation of knowledge-based approaches to safety education.

The findings underscore the importance of shifting towards behaviourally oriented, ecologically valid training approaches that promote the generalisation of safety skills across real-life contexts. Without such interventions, individuals with intellectual disabilities remain at significant risk of victimisation despite possessing theoretical knowledge.

Ultimately, enhancing personal safety skills is not only a matter of education but a fundamental issue of protection, autonomy and human rights.