At-risk Youth in Honduras: A Psychometric Evaluation of the Instrumento de Medicion de Comportamientos (IMC)

Abstract

Prior research highlights the importance of developmental domains—community, family, school, peer, and individual—in shaping youth problem behaviors such as delinquency, drug use, and violence. However, few instruments measuring risk and protective factors have been psychometrically evaluated in developing countries, particularly in Central America. This study examines the psychometric properties of the Instrumento de Medicion de Comportamientos (IMC), a prevention-oriented survey assessing risk and protective factors, using a sample of Honduran youth. We evaluated the IMC’s internal consistency, construct validity, concurrent validity, and measurement invariance. Findings indicate the IMC demonstrates acceptable reliability and validity in measuring key risk and protective constructs in Honduras. These findings support the IMC as a tool for identifying at-risk youth, informing prevention and early intervention efforts, and extending the psychometric evaluation of youth risk and protective factor instruments to Latin America, highlighting the importance of locally developed instruments for evidence-based crime prevention and youth violence interventions.

Keywords

validity risk factors protective factors risk assessment delinquency

Introduction

Prior literature examining risk factors associated with problem behaviors among youths typically utilizes developmental domains, including community, family, school, peer, and individual (Arthur et al., 2007), to better understand the contexts and environments that predict youth problem behaviors like delinquency, drug use, and gang membership. Understanding how risk factors influence problem behaviors across these developmental domains is beneficial because many scales provide nuanced measurements, enabling clearer assessment of their relative influence. Howell and Egley (2005) suggest that longitudinal studies confirm that risk factors in any of these domains can increase future problem behaviors.

Research has focused on developing and implementing sophisticated risk and protective factors measures using robust psychometric methods, including validation in low- and middle-income countries (Murray et al., 2018). While some risk assessment tools are currently used in Latin America, they could be improved through further research and validation within those countries. From a development perspective, locally developed and validated instruments better support Latin American capacity for evidence gathering, informing policy, promoting information sharing, and guiding prevention and intervention decisions, particularly in high-crime areas.

Certainly, the Honduran context is distinctive. Honduras consistently ranks among the most violent nations globally, with a homicide rate of 89 per 100,000 inhabitants in 2011 (Landa-Blanco et al., 2020). Although progress has been made toward reducing the number of homicides, the rate remains considerably high at 38 per 100,000 inhabitants, approximately 2.7 times the regional average for Latin America as a whole (United Nations Office on Drugs and Crime, 2023). Youth in Honduras are exposed to violence and insecurity without adequate community support or protection (De Jesus & Hernandes, 2019). Police protection remains inconsistent and lacks legitimacy; gangs can function as quasi-government actors. School safety and related environments vary significantly; moreover, family systems are characterized by extended and multi-generational households facing severe economic pressures (Frank-Vitale & Hoffnung-Garskof, 2025).

Because most youth risk assessment tools have been developed and validated in English-speaking, developed countries, direct translation into Spanish does not ensure that the items and scales will possess psychometric reliability and validity. If translated instruments yield unreliable or invalid risk assessments, youth may be misclassified. This could result in high-risk youth being misclassified as low risk, and low-risk youth being misclassified as high risk, ultimately compromising the efficiency and effectiveness of the programs. Collectively, these issues underscore the critical need for precise risk assessment in Honduras to inform preventive and intervention strategies that enhance both efficacy and efficiency.

The present study aims to examine the psychometric properties of the youth risk assessment tool known as Instrumento de Medicion de Comportamientos (IMC) (in English: Behavior Measurement Instrument), which primarily consists of items from the Communities that Care Youth Survey (CTCYS) and has been validated in the U.S. (Arthur et al., 2007), Australia (Beyers et al., 2004), and parts of the Caribbean (Katz & Fox, 2010). However, it has not yet been validated in Central America, including Honduras. A locally validated tool would enable service providers and intervention specialists to identify the factors that place youth in Honduras at risk or protect them from delinquency, violence, and other problem behaviors.

Literature Review

Significant effort has been invested in understanding and preventing adolescent problem behaviors, including delinquency, drug use, gang involvement, and truancy. A promising development in this body of knowledge is the risk and protective model, designed to identify specific factors associated with an increased or decreased likelihood of such behaviors. Similar to the epidemiological approach, it examines patterns of vulnerability and susceptibility linked to risk and protective factors (Haggerty & Mrazek, 1994). Widely used to address health, social, and mental issues (e.g., Shaffer & Yates, 2010), the model has significant policy implications by aiming to prevent problem behaviors through “eliminating, reducing, or mitigating their precursors” (Hawkins et al., 1992, p. 65). While the concept is straightforward in principle, it becomes complicated in practice due to numerous potential risk and protective factors that may influence problem behaviors.

Notably, previous research underscores that protective factors are conceptually distinct from risk factors and may function as separate promotive influences or buffering mechanisms that mitigate the effects of risk, rather than merely representing the absence of risk (Farrington et al., 2012). Recent advances in prevention science show a shift from focusing solely on risk reduction to strengthening protective factors across different domains, which shield youth from problematic behaviors. Vincent et al. (2025) state that protective factors are not merely the opposite of risk factors; they are shaped by social, family, educational, and community contexts and have effects that vary and accumulate. Despite their importance in evidence-based prevention, measurement approaches remain inconsistent, especially in high-risk settings like Honduras, where culturally appropriate and reliable measurement tools are needed to assess youth strengths and guide prevention and intervention strategies.

Research on delinquency and related problem behaviors emphasizes the interplay of risk and protective factors operating across multiple domains, including individual, family, peer, school, and community context. Risk factors, such as impulsivity, exposure to delinquent peers, or neighborhood violence, increase the likelihood of negative developmental outcomes, whereas protective factors, including strong family bonds, prosocial peer relationships, and school engagement, attenuate the impact of these risks (Fairnington, 2004; Hawkins et al., 1992). Developmental research underscores that these factors function interactively: cumulative risk heightens vulnerability to maladaptive outcomes, while protective factors can mitigate these effects, even in high-risk environments (Loeber & Farrington, 2000; Masten & Coatsworth, 1998). This framework informs a foundation for preventive interventions by highlighting ways to strengthen protective factors and reduce exposure to risks.

The Social Development Model (SDM) offers an integrative theoretical framework, synthesizing constructs from social control (Hirschi, 1969), social learning (Akers, 1973), and differential association theories (Sutherland, 1973) to explain the development of prosocial and antisocial behavior over time (Catalano & Hawkins, 1996). The SDM posits that opportunities for involvement, acquisition of relevant skills, and recognition or rewards within social units, such as family, school, peers, and community, facilitate bonding, which in turn promotes internalization of the normative standards, leading to behavioral outcomes. Previous research has validated the developmental risk and protective processes of the SDM with diverse populations (Roosa et al., 2011). Overall, integrating the SDM within the risk and protective factor framework underscores the mechanisms by which social environments shape behavior, highlighting that prevention efforts should focus not only on reducing exposure to risk factors but also on strengthening social bonds and promoting prosocial competencies. This perspective has become a cornerstone in developmental criminology and prevention science, providing both theoretical clarity and practical guidance for intervention.

As cross-national research on the relationship between risk and protective factors and problem behavior expands, it is likely to yield insights into how these correlates function in various settings (e.g., Beyers et al., 2004; Ohene et al., 2005). Identifying and understanding risk and protective factors is especially crucial in developing nations, where progress hinges on optimizing the use of limited resources to maximize the chances of achieving positive outcomes (Blum & Ireland, 2004). Therefore, it is critical to determine whether the measurement properties of risk and protective factors assessed by survey instruments are applicable outside of the United States, particularly in developing nations where the nature of risk and protection may differ and where evidence indicates that prevention efforts are necessary.

The Communities that Care (CTC) is a widely used prevention framework across the United States. It is a comprehensive, evidence-based approach grounded in the social development model that aims to reduce youth problem behaviors by addressing risk and protective factors through coordinated community action (Catalano & Hawkins, 1996; Hawkins et al., 2017). As part of this framework, the Communities That Care Youth Survey (CTCYS) is a standardized self-report instrument that measures empirically derived risk and protective factors across individual/peer, family, school, and community domains. The individual/peer domain captures constructs like peer drug use and sensation seeking; the family domain assesses factors like poor family management and family conflict; the school domain includes academic failure and low commitment to school; and the community domain measures factors like community disorganization and neighborhood attachment. The survey serves multiple purposes, such as identifying suitable prevention programs and understanding the factors linked to drug use, delinquency, and other adolescent problem behaviors. It is beginning to be used in other countries, including Canada (Flynn, 2008), Australia (Beyers et al., 2004), the Netherlands (Jonkman et al., 2015), and the United Kingdom (Fairnington, 2004).

Unlike many commonly used risk and needs assessment tools, the CTCYS and IMC are grounded in a population-level, developmental prevention orientation informed by the SDM. The IMC assesses malleable risk and protective factors across multiple domains to guide community-level prevention planning rather than predicting individual outcomes (Arthur et al., 2002). In contrast, tools like the Youth Level of Service/Case Management Inventory (YLS/CMI) are grounded in the Risk-Need-Responsivity model and emphasize individual criminogenic risk and need to inform case management (Andrews et al., 2011). In addition, trauma-focused instruments like the Adverse Childhood Experiences (ACEs) questionnaire measure cumulative childhood adversity to predict health and behavioral outcomes (Felitti et al., 1998), while substance-use screeners like CRAFFT (n.d.) target adolescent substance misuse in a clinical context. Although these tools are evidence-based and widely used, they differ from the CTCYS/IMC in theoretical orientation, scope, and intended use, supporting population-level prevention strategies rather than individual diagnosis or justice decisions.

While studies have examined and validated CTCYS scales in the United States, little is known about the properties of these scales in other regions, especially in high-crime developing nations. The cultural, structural, and political differences between countries make it crucial to test measures that have been developed and validated in other contexts. Glaser et al. (2005) emphasized that using the instrument is “contingent on a valid assessment of levels of risk and protection within a community” (p. 75). They further note that for prevention programs to be effective, the measurement tools used to determine the extent and nature of risk and protection must be valid across geographic areas and subpopulations.

Despite increased use of the CTCYS beyond the United States, limited research has assessed its reliability and validity in non-US contexts, especially in developing nations where evidence-based tools are vital. International evidence regarding its psychometric performance remains mixed and domain-specific. In particular, the school and community domains tend to exhibit greater psychometric instability. School risk factors–like school attachment and academic engagement–are susceptible to differences in educational systems, school organization, and normative expectations regarding teacher-student relationships, attendance, and disciplinary practices (Thapa et al., 2013). Desa et al. (2019) examined measurement invariance in international large-scale assessments, including over 5,300 schools from 38 countries in Europe, Latin America, and the Asia-Pacific region. They found that school-and institution-related constructs (e.g., attitudes toward school) rarely achieve full measurement invariance across countries due to systematic differences in educational systems and normative contexts. Similarly, community-level constructs, including neighborhood disorganization and exposure to violence, often vary in meaning across sociopolitical contexts, particularly in settings where informal governance, gang presence, or limited state capacity shape daily life (Koonings & Kruijt, 2007). Maguire et al. (2011) reported weak construct validity for most community-level factors in Trinidad and Tobago, with limited concurrent validity even after scale revision. A recent systematic review by Thurow et al. (2021) further noted that while most CTCYS evaluations outside the United States demonstrated generally acceptable psychometric properties, these were weaker than those observed in U.S.-based studies and U.S. subpopulation adaptations. Taken together, CTCYS performs adequately in many international settings, but risk and protective factors–especially those embedded in school and community contexts–may not function equivalently across nations, highlighting the importance of empirical validation in places such as Honduras.

The present study examines the psychometric properties of the risk and protective factor scales included in the IMC, which primarily consists of items from the CTCYS, among a sample of Honduran youth. Specifically, this research evaluates the IMC’s internal consistency, construct validity, concurrent validity, and measurement invariance, providing a comprehensive assessment of the tool’s psychometric robustness. The study seeks to extend prior research on risk and protective factors for youth by assessing whether the scales are reliable and accurately measure the constructs they are intended to measure in a region with a distinct cultural, social, and historical context. This work contributes to the broader application of evidence-based tools across diverse international settings, highlighting potential adaptation and considerations necessary for cross-cultural research and intervention.

Data and Methods

The Research Setting

Over the past several years, Honduras has experienced some of the highest homicide rates in the world. In 2011, this rate was 86.5 homicides per 100,000 people. While the homicide problem has significantly decreased, as recently as 2018, the country still ranked among the most violent globally, with 31.1 homicides per 100,000 people (Insight Crime, 2024). In 2014, in response to rising violence, USAID completed the Honduras Country Development Cooperation Strategy, which called for US developmental resources to focus on the densely populated urban communities most at risk for violence. This initiative included 14 communities within the two largest municipalities: Tegucigalpa, the nation’s capital and largest city, and San Pedro Sula, the industrial center and second largest city in Honduras. It also encompassed three other municipalities: Choloma, La Ceiba, and Tela. Each of these five municipalities had homicide rates that exceeded the national average.

Data

Data for the current study were collected as part of a larger project in Honduras, Proponte Más, which aims to reduce risk factors and enhance protective factors among at-risk youth living in high-risk communities (citation removed for review purposes). This study uses data from the Instrumento de Medicion de Comportamientos (IMC) data, which was administered to youth living in fourteen communities across five municipalities in Honduras, including Distrito Central (Tegucigalpa), San Pedro Sula, La Ceiba, Choloma, and Tela, who were referred to the Proponte Más program.¹ It was conducted prior to program entry to assess risk and protective factors for determining program eligibility and again following treatment to evaluate change over time. Youth with four or more identified risk factors assessed by the pre-treatment IMC were eligible to participate in the program. Eligible youth and their families received six months of counseling and structured activities delivered by certified family counselors to promote engagement and behavioral improvement. Only pre-treatment IMC data were used in the present analyses to enhance generalizability. IMC largely consists of questions from the Communities that Care Youth Survey (CTCYS).² It includes items that measure socio-demographic characteristics and 173 items assessing 38 risk and protective factors across four domains (e.g., community, school, family, peer/individual) (see Supplemental Appendix A for descriptive statistics; online supplement, available in the online version of this article).

Sample

A total of 4,574 youths were referred to the program in 2017. Written consent was obtained from the parents or caregivers of 4,495 of these youths, and all 4,495 youths completed the IMC for 30 days. Most youth were referred by their parent(s) or guardians (54.7%) or their school (29.7%). The remaining referrals came from a USAID-sponsored outreach center (3.0%), other family members (2.9%), a program or institution (2.7%), a church (1%), self-referrals (0.4%), counselors or advisors (0.2%), or another source (5.3%). Youth ranged in age from 8 to 17, with a mean age of 12.33 (SD = 2.6). Nearly 60 percent were male, and most youth were in school (88.7%) and were born in urban areas (93.0%).

Measures

The community domain includes five risk factor scales (i.e., transitions and mobility, low neighborhood attachment, community disorganization, laws and norms favorable to drug use, and perceived availability of drugs) and two protective factor scales (i.e., opportunities for prosocial involvement and rewards for prosocial involvement). The family domain comprises six risk factor scales (i.e., family history of antisocial behavior, parental attitudes favorable toward drug use, poor family management, family conflict, weak parental supervision, and family gang influence) and three protective factor scales (i.e., attachment, opportunities for prosocial involvement, and rewards for prosocial involvement). The school domain features two risk factor scales (i.e., academic failure and low commitment to school) and two protective factor scales (i.e., opportunities for prosocial involvement and rewards for prosocial involvement). The peer/individual domain comprises 14 risk factor scales (i.e., rebelliousness, rewards for antisocial involvement, favorable attitudes toward drug use, favorable attitudes toward antisocial behavior, perceived risks of drug use, friends’ drug use, interactions with antisocial peers, intentions to use, antisocial tendencies, critical life events, impulsive risk-taking, neutralization of guilt, negative peer influence, and peer delinquency) and four protective factor scales (i.e., belief in the moral order, rewards for prosocial involvement, interactions with prosocial peers, and social skills). Detailed item-level information and coding schemes for each risk and protective factor are provided as supplementary material due to space limitations (see Supplemental Appendix A, available in the online version of this article).

In addition, the IMC gathers information about participants’ problem behaviors over the previous 6 months. Eighteen items measure six types of problem behaviors, including violent behavior, property crime, gang involvement, alcohol/drug use, drug selling, and carrying weapons (see Supplemental Appendix B, available in the online version of this article). Each outcome variable is scored as “1” if a respondent reported engaging in any delinquent behavior and “0” if none were endorsed. Three items evaluate violent behavior (e.g., in the last 6 months, have you hit someone with the purpose of hurting him or her?), four items assess property crime (e.g., in the past 6 months, have you purposely damaged or destroyed things that do not belong to you?), three items gauge gang involvement (e.g., in the past 6 months, have you been a member of a gang?), three items measure alcohol and drug use (e.g., in the past, have you used marijuana or any other illegal drugs?), two items measure drug sales (e.g., in the past 6 months, have you sold or helped sell marijuana or other illegal drugs?), and two items measure weapon’s carrying (e.g., in the past 6 months, have you carried a concealed weapon for protection?).

Analytic Strategy

We conducted several analyses to validate the Honduran IMC.³ First, we examined the reliability of each sub-factor. One common method to test reliability is internal consistency, which reflects the extent to which all items on the same scale consistently measure the same construct or concept. In this paper, we used the omega (ω) coefficient for assessing internal consistency (McDonald, 1999).⁴

Second, we examined the measurement properties of the IMC subcomponents through a series of CFA models, which were conducted using Maximum Likelihood (ML) estimation, as the sample size was large and preliminary analyses indicated no problematic skewness or kurtosis in the items. Model fit assessment relied on several indices, including the chi-square statistic, root mean square error of approximation (RMSEA), comparative fit index (CFI), and Tucker–Lewis index (TLI). We employed standard acceptability ranges to evaluate model fit (RMSEA ≤ .06, CFI ≥ .95, TLI ≥ .95; Hu & Bentler, 1999). Notably, a small number of items (5 out of 173) had loadings on the latent factor that fell below the common threshold of .3 (Tabachnick et al., 2007). We chose to retain those items in the final CFA model because they address important aspects of the construct and enhance the scale’s content validity.

Third, we examined concurrent validity by examining interrelations between risk and protective factors and various delinquent behaviors, including violence, property crime, gang involvement, drug use, drug selling, and weapon carrying. In the analyses, protective factors were reverse-coded to align with risk factors. It is for analytic consistency so that higher values across all constructs reflect greater risk. Given that the items exhibited inconsistent scale ranges, we utilized Percent of Maximum Possible (POMP) scores, which represent the position of the response on the scale as a percentage of the maximum achievable score (Cohen et al., 1999).⁵ After processing the POMP scores for risk and protective factor items, point-biserial correlation coefficients were calculated to assess the strength of the association between delinquent behavior outcomes and risk and protective factor scores.

Last, we investigated the generalizability of the IMC tool through measurement invariance (MI) testing to confirm that the CFA models are consistent across different groups of people by estimating changes in the goodness of fit indices between models. MI is a technique that helps to examine whether the same unobserved variable is being measured across multiple groups within a CFA framework (Vandenberg & Lance, 2000). If the MI assumptions are tenable, then the IMC tool is generalizable to different groups. The procedure involves a series of analyses that evaluate the different types of invariance for each domain.⁶ In our MI analysis, we focused on testing measurement invariance between females and males and between the school and non-school youth.⁷ First, a baseline (configural) model was fit to the data. In this model, all scales are defined by the same subset of indicators between groups (i.e., the factor patterns are the same). Second, invariance factor loadings were examined to determine whether each factor loading was similar across the samples (metric model). Third, invariance for threshold was examined by adding the constraint of equal thresholds between groups (scalar model). Last, invariance for factor variance-covariances was tested by adding the constraint of equal variance-covariance values across the groups (residual model).⁸

Results

Reliability

Table 1 presents the coefficient ω and α for each risk and protective factor. Overall, most of the α estimates for the IMC are .70 or above, indicating acceptable internal consistency (McNeish, 2018; α values of .80 indicate good reliability and values of .90 or above indicate excellent reliability), which means that the IMC items consistently measure the same underlying dimension. For ω reliability, there is no universally acceptable threshold for adequate levels, but some researchers suggest that ω should exceed .50 at a minimum, with .75 preferable (Watkins, 2017). We have ω values based on CFA models, with four 2-item scales (i.e., low neighborhood attachment, academic failure, family gang influence, and intentions to use drugs) lacking ω because CFA require at least three items. To supplement this, we added Cronbach’s α, which is also available for 2-item scales. Note that “alpha is very much a function of the number of items in a scale” (Cortina, 1993, p. 102). Therefore, shorter scales (2–5 items) tend to have lower α values, partly explaining low scores.

Table 1:

Reliability of Risk and Protective Factors, Honduras (n = 4,495)

Factors	ω	α
Community risk factors
Transitions and Mobility	0.830	0.573
Low Neighborhood Attachment	NA	0.394
Community Disorganization	0.822	0.758
Laws and Norms Favorable to Drug Use	0.862	0.637
Perceived Availability of Drugs	0.910	0.818
Community protective factors
Opportunities for Prosocial Involvement	0.749	0.579
Rewards for Prosocial Involvement	0.888	0.788
School risk factor ^a
Academic Failure	NA	0.544
Low Commitment to School	0.709	0.614
School protective factor ^a
Opportunities for Prosocial Involvement	0.643	0.473
Rewards for Prosocial Involvement	0.793	0.691
Family risk factors
Family History of Antisocial Behavior	0.859	0.699
Parental Attitudes Favorable Toward Drug Use	0.896	0.768
Poor Family Management	0.851	0.712
Family Conflict	0.849	0.750
Weak Parental Supervision	0.787	0.671
Family Gang Influence	NA	0.279
Family protective factors
Attachment	0.908	0.805
Opportunities for Prosocial Involvement	0.801	0.633
Rewards for Prosocial Involvement	0.669	0.570
Peer/individual risk factors
Rebelliousness	0.612	0.536
Rewards for Antisocial Involvement	0.908	0.806
Favorable Attitudes Toward Drug Use	0.832	0.719
Favorable Attitudes Toward Antisocial Behavior	0.893	0.837
Perceived Risks of Drug Use	0.839	0.754
Friends’ Use of Drugs	0.896	0.737
Interaction with Antisocial Peer	0.802	0.621
Intentions to Use Drug	NA	0.548
Antisocial Tendencies	0.655	0.566
Critical Life Events	0.589	0.405
Impulsive Risk Taking	0.803	0.753
Neutralization of Guilt	0.744	0.717
Negative Peer Influence	0.820	0.745
Peer Delinquency	0.860	0.719
Peer/individual protective factors
Belief in the Moral Order	0.665	0.449
Rewards for Prosocial Involvement	0.691	0.621
Interaction with Prosocial Peers	0.874	0.809
Social Skills	0.422	0.386

Note. Four 2-item scales do not have an ω value because ω is calculated using CFA results and CFA models require at least three items.

School domain only includes the school sample.

For community risk and protective factors, all the subscales show between an excellent and an acceptable level of internal consistency, except for the low neighborhood attachment scale. For school risk and protective factors, low commitment to school and rewards for prosocial involvement scales show acceptable internal consistency; however, academic failure and opportunities for prosocial involvement indicate poor internal consistency. Concerning family risk and protective factors, all the scales indicate between an excellent and acceptable level of reliability, except family gang influence and rewards for prosocial involvement. The rewards for prosocial involvement scale indicate questionable internal consistency. For the peer/individual domain, all peer/individual risk factors indicate good internal consistency except rebelliousness, intentions to use drugs, antisocial tendencies, and critical life events. Among peer/individual protective factors, the interaction with prosocial peers scale shows a good level of reliability; however, the other three scales, including belief in the moral order, rewards for prosocial involvement, and social skill scales, indicate questionable or poor internal consistency.

Construct Validity

Overall, the majority of 173 items with 38 subcomponents retained good factor validity with good model fits. Supplemental Figures S1 through S4, available in the online version of this article, present the final measurement models containing the risk and protective factors by domain.⁹ Model fits of each risk and protective factor are presented with model fit indices in Table 2. The measurement models containing community risk and protective factors have standardized factor loadings ranging from .16 to .98. Item CRc3 in community disorganization (λ = .24) and two items—CRd4 and CRd5—in laws and norms favorable to drug use show lower factor loadings (λ = .16, λ = .22, respectively).

Table 2:

Model Fit Indices by Each Risk and Protective Factor, Honduras (n = 4,495)

Risk and protective factors	χ²	df	RMSEA	CFI	TLI
Community risk factors
Transitions and Mobility^a	14.528	1	0.055	0.994	0.967
Low Neighborhood Attachment	NA
Community Disorganization^a	991.601	31	0.083	0.940	0.912
Laws and Norms Favorable to Drug Use	13.195	4	0.023	0.999	0.998
Perceived Availability of Drugs	23.893	2	0.049	0.999	0.997
Community protective factors
Opportunities for Prosocial Involvement	35.996	5	0.037	0.989	0.978
Rewards for Prosocial Involvement	.000	0	0.000	1.000	1.000
School risk factors^b
Academic Failure	--------------------------NA--------------------------
Low Commitment to School	189.971	14	0.056	0.954	0.932
School protective factors ^b
Opportunities for Prosocial Involvement	18.149	5	0.026	0.990	0.981
Rewards for Prosocial Involvement	42.486	2	0.071	0.995	0.984
Family risk factors
Family History of Antisocial Behavior^a	722.589	33	0.068	0.930	0.905
Parental Attitudes Favorable Toward Drug Use	123.002	5	0.072	0.988	0.975
Poor Family Management	563.142	20	0.078	0.937	0.912
Family Conflict	.000	0	0.000	1.000	1.000
Weak Parental Supervision	76.841	5	0.056	0.985	0.971
Family Gang Influence	--------------------------NA--------------------------
Family protective factors
Attachment^a	1116.232	19	0.113	0.960	0.941
Opportunities for Prosocial Involvement	.000	0	0.000	1.000	1.000
Rewards for Prosocial Involvement^a	4.984	1	0.030	0.999	0.993
Peer/individual risk factors
Rebelliousness	.000	0	0.000	1.000	1.000
Rewards for Antisocial Involvement:	.000	0	0.000	1.000	1.000
Favorable Attitudes Toward Drug Use	.000	0	0.000	1.000	1.000
Favorable Attitudes Toward Antisocial Behavior	58.233	5	0.049	0.996	0.992
Perceived Risks of Drug Use	.000	0	0.000	1.000	1.000
Friends’ Use of Drugs	.000	0	0.000	1.000	1.000
Interaction with Antisocial Peer	56.237	2	0.078	0.981	0.942
Intentions to Use	--------------------------NA--------------------------
Antisocial Tendencies^a	183.795	12	0.056	0.948	0.908
Critical Life Events^a	44.508	12	0.025	0.967	0.942
Impulsive Risk Taking	24.512	2	0.050	0.998	0.994
Neutralization of Guilt^a	33.244	7	0.029	0.999	0.997
Negative Peer Influence	.000	0	0.000	1.000	1.000
Peer Delinquency^a	5.519	4	0.009	1.000	0.999
Peer/individual protective factors
Belief in the Moral Order	7.406	2	0.025	0.996	0.989
Rewards for Prosocial Involvement	13.348	2	0.036	0.997	0.991
Interaction with Prosocial Peers	120.336	5	0.072	0.995	0.989
Social Skills	.000	0	0.000	1.000	1.000

Note. Four 2-item scales do not have model fit indices because CFA models require at least three items.

We applied modification indices (i.e., adding pairs of residual correlations) to improve the model fit. These modification indices are presented in APPENDIX C.

School domain only includes the school sample.

Several modification indices were suggested from the models, and model adjustments were implemented when supported by clear substantive or theoretical justification (see Supplemental Appendix C, available in the online version of this article). Specifically, correlation residuals were added between items that exhibited substantial semantic overlap, shared referents, or closely related behavioral content, consistent with recommended practices in CFA. For community risk factors, residual covariance was added between items assessing Transitions and Mobility—changing homes and changing schools—given that both capture instability in residential and educational contexts and are often experienced concurrently during adolescence. Within Community Disorganization, residual correlations were specified among items assessing neighborhood crime/drug selling, fight, graffiti, and abandoned buildings, as these indicators reflect overlapping manifestations of neighborhood disorder and social disorganization. Similarly, for Laws and Norms Favorable to Drug Use, residual covariance was added between perceptions of adult norms regarding adolescent marijuana and alcohol use, reflecting a shared normative climate surrounding substance use rather than a distinct construct.

For family risk factors, residual correlations were added among items assessing sibling alcohol use, marijuana use, and cigarette smoking, as these items reference the same family members and reflect a shared underlying exposure to antisocial behavior within the household. Among family protective factors, a residual covariance was added between items assessing closeness with and emotional communication with one’s father, as well as between items assessing parental praise and recognition of positive behavior. In both cases, the paired items capture closely related dimensions of parental attachment and reinforcement processes.

For peer and individual risk factors, correlated residuals were specified between items assessing kindness, respect for others’ feelings, and obedience, as these items reflect overlapping prosocial dispositions within the Antisocial Tendencies scale. Within Critical Life Events, residual correlations were added between items capturing academically and socially disruptive experiences, as well as between items assessing exposure to serious injury or death from illness versus from violence, reflecting shared emotional and contextual stressors. To neutralize guilt, residual covariance was added between items that justify theft under similar moral rationales and between items that justify retaliatory or defensive violence, reflecting a common cognitive mechanism of moral disengagement. Finally, within Peer Delinquency, residual covariance was added between items assessing friends’ theft and violent behavior, as both capture involvement in a delinquent peer network.

After adjusting the models based on modification indices, each model of community risk and protective factors reaches an adequate range of fit (see Table 2). The measurement models contain the school risk and protective factors. It presents standardized factor loadings ranging from 0.41 to 0.93. The model fit indices also suggest that each model fits the data well, with the adjustment based on modification indices. The measurement models for the family risk and protective factors show standardized factor loadings ranging from 0.34 to 0.87. After adjusting the models based on modification indices, each model reaches an appropriate range of fit. Supplemental Figure S4, available in the online version of this article, includes the measurement models examining peer/individual risk and protective factors. The standardized factor loadings range from 0.19 to 0.96. Item IRj6 in critical life events (λ = 0.19), and IPo1 in belief in the moral order (λ = 0.28) show lower factor loadings. After the model adjustment based on modification indices, the models fit the data well.

Concurrent Validity

Table 3 shows the point biserial correlation coefficients for the relationships between risk and protective factor scales and six problem behaviors. We found that most of the risk factors had positive correlations with each delinquency outcome, ranging between .01 and 0.48, with the expected direction. All 27 risk factors exhibit small effects on violence, property crime, and carrying weapons (r < .30). For gang involvement, 24 risk factors show small effects, while three demonstrate medium effects (.30 < r < .70). Regarding drug use, 23 risk factors have small effects, and four have medium effects. For drug selling, 25 risk factors show small effects, and two show medium effects. Some protective factors, which were reverse-coded to align directionally with risk factors, were negatively associated with delinquent behaviors; however, these associations were very weak (small effect), indicating that higher levels of these protective factors only slightly increased delinquent behaviors. These protective factors are: (1) community opportunities for prosocial involvement with property crime (r = −.03), with gang involvement (r = −.01), and with drug selling (r = −.01), (2) community rewards for prosocial involvement with all delinquent behaviors (r ranged between −.01 and −.10), (3) school opportunities for prosocial involvement with drug use (r = −.01), and (4) interaction with prosocial peers with property crime (r = −.01).

Table 3:

Concurrent Validity Evidence of Seven Problem Behaviors to Youth Risk/Protective Factors (Point Biserial Correlation), Honduras (n = 4,495)

Factors	Violent behavior	Property crime	Gang involvement	Drug use	Drug selling	Carry weapon
Community risk factors
Transitions and Mobility	0.05	0.09	0.07	0.06	0.05	0.03
Low Neighborhood Attachment	0.07	0.08	0.05	0.04	0.03	0.05
Community Disorganization	0.09	0.10	0.10	0.05	0.05	0.03
Laws and Norms Favorable to Drug Use	0.13	0.12	0.13	0.19	0.09	0.08
Perceived Availability of Drugs	0.20	0.23	0.26	0.33	0.17	0.18
Community protective factors
Opportunities for Prosocial Involvement	0.02	-0.03	-0.01	0.04	-0.01	-0.01
Rewards for Prosocial Involvement	-0.06	-0.03	-0.03	-0.10	-0.01	-0.04
School risk factors ^a
Academic Failure	0.11	0.13	0.06	0.18	0.08	0.09
Low Commitment to School	0.20	0.20	0.14	0.25	0.14	0.14
School protective factors ^a
Opportunities for Prosocial Involvement	0.05	0.04	0.01	-0.01	0.02	0.03
Rewards for Prosocial Involvement	0.08	0.09	0.05	0.15	0.05	0.09
Family risk factors
Family History of Antisocial Behavior	0.25	0.28	0.29	0.34	0.24	0.19
Parental Attitudes Favorable Toward Drug Use	0.12	0.17	0.17	0.15	0.11	0.07
Poor Family Management	0.16	0.16	0.18	0.22	0.13	0.11
Family Conflict	0.20	0.23	0.17	0.21	0.12	0.11
Weak Parental Supervision	0.16	0.21	0.15	0.16	0.11	0.08
Family Gang Influence	0.19	0.22	0.30	0.29	0.23	0.17
Family protective factors
Attachment	0.18	0.18	0.18	0.29	0.12	0.13
Opportunities for Prosocial Involvement	0.14	0.10	0.13	0.20	0.07	0.08
Rewards for Prosocial Involvement	0.16	0.14	0.15	0.24	0.10	0.11
Peer/individual risk factors
Rebelliousness	0.19	0.20	0.16	0.17	0.14	0.07
Rewards for Antisocial Involvement	0.13	0.13	0.17	0.23	0.17	0.14
Favorable Attitudes Towards Drug Use	0.08	0.08	0.14	0.15	0.11	0.09
Favorable Attitudes Towards Antisocial Behavior	0.09	0.10	0.12	0.08	0.09	0.08
Perceived Risks of Drug Use	0.09	0.12	0.11	0.08	0.11	0.06
Friends’ Use of Drugs	0.23	0.25	0.38	0.48	0.31	0.23
Interaction with Antisocial Peer	0.23	0.25	0.31	0.24	0.26	0.22
Intentions to Use	0.22	0.22	0.26	0.41	0.22	0.20
Antisocial Tendencies	0.23	0.25	0.15	0.15	0.09	0.07
Critical Life Events	0.24	0.27	0.26	0.30	0.18	0.19
Impulsive Risk Taking	0.25	0.29	0.24	0.25	0.15	0.12
Neutralization of Guilt	0.23	0.27	0.20	0.20	0.14	0.12
Negative Peer Influence	0.12	0.17	0.19	0.15	0.13	0.07
Peer Delinquency	0.27	0.29	0.37	0.30	0.35	0.26
Peer/individual protective factors
Belief in the Moral Order	0.19	0.22	0.19	0.21	0.14	0.12
Rewards for Prosocial Involvement	0.08	0.09	0.05	0.06	0.06	0.04
Interaction with Prosocial Peers	0.03	-0.01	0.02	0.05	0.05	0.04
Social Skills	0.17	0.16	0.18	0.24	0.18	0.16

School domain only includes the school sample.

Measurement Invariance

Table 4 presents the results of our measurement invariance tests, which assessed the generalizability of the risk and protective scales across gender and school status. The results of the first set of analyses suggest that all the risk and protective factor scales are valid for both males and females. Specifically, first, the model fit indices for the baseline models are acceptable across four domains. This supports the configural invariance of all four domains in male and female subsamples. It suggests that male and female youth in the IMC seem to have the same basic conceptualization of each risk and protective factor. Second, given that configural invariance was supported, we examined metric invariance to examine the consistency of each item’s loading between male and female subsamples. The results of model comparison between configural invariance model and the loading invariance model supported metric invariance (range of ∆RMSEA = .001 – .003, range of ∆CFI = .002 – .014, range of ∆TLI = .006 – .017), indicating that the item factor loadings are consistent for male and female subsamples. Third, given support for metric invariance, scalar invariance was tested and supported range of ∆RMSEA = .000 – .002, range of ∆CFI = .002 – .003, range of ∆TLI = .002 – .007). This indicates that the measurement intercepts of each risk and protective scale are consistent across male and female respondents. In other words, the result establishes a common zero point on the factors for males and females, thus allowing meaningful comparisons of the latent means. Last, given that scalar invariance was supported, we examined residual invariance. Changes in model fit indices indicate support for residual invariance (∆RMSEA = .000, ∆CFI = .000, ∆TLI = .000), suggesting that residuals of each risk and protective scale are invariant across male and female youth. Overall, the results of all of the above invariance tests suggest that risk and protective factors in each domain function equivalently well for both males and females in the sample.

Table 4:

Fit Indicators Taken From CFA and Invariance Analyses, Honduras (n = 4,495)

Invariance models	Overall fit indices					Model comparison	Comparative fit indices
	χ2	df	RMSEA	CFI	TLI	Model comparison	∆χ2	∆df	∆RMSEA	∆CFI	∆TLI
Gender invariance (male, n = 2,689 vs. female, n = 1,806)
Community domain
M0: Configural Invariance Model (Baseline model)	7601.023*	946	0.056	0.889	0.877
M1: Metrics invariance Model	7453.751*	972	0.054	0.892	0.883	M1 vs. M0	46.288	26	-0.002	0.003	0.006
M2: Scalar invariance Model	7640.784*	1048	0.053	0.890	0.890	M2 vs. M1	183.075*	76	-0.001	-0.002	0.007
M3: Residual Invariance Model	7640.784*	1048	0.053	0.890	0.890	M3 vs. M2	104.596*	0	0.000	0.000	0.000
School domain
M0: Configural Invariance Model (Baseline model)	824.308*	258	0.033	0.965	0.958
M1: Metrics invariance Model	802.313*	276	0.031	0.967	0.964	M1 vs. M0	38.537	18	-0.002	0.002	0.006
M2: Scalar invariance Model	896.370*	332	0.029	0.965	0.968	M2 vs. M1	83.105	56	-0.002	-0.002	0.004
M3: Residual Invariance Model	896.370*	332	0.029	0.965	0.968	M3 vs. M2	28.496	0	0.000	0.000	0.000
Family domain
M0: Configural Invariance Model (Baseline model)	13573.168*	2088	0.049	0.870	0.860
M1: Metrics invariance Model	12380.381*	2127	0.046	0.884	0.877	M1 vs. M0	154.915*	39	-0.003	0.014	0.017
M2: Scalar invariance Model	12813.613*	2265	0.046	0.881	0.881	M2 vs. M1	558.178*	138	0.000	-0.003	0.004
M3: Residual Invariance Model	12813.613*	2265	0.046	0.881	0.881	M3 vs. M2	192.368*	0	0.000	0.000	0.000
Peer/Individual domain
M0: Configural Invariance Model (Baseline model)	18102.818*	4948	0.034	0.902	0.892
M1: Metrics invariance Model	17177.006*	5004	0.033	0.909	0.902	M1 vs. M0	108.343*	56	-0.001	0.007	0.010
M2: Scalar invariance Model	17737.036*	5238	0.033	0.906	0.904	M2 vs. M1	883.448*	234	0.000	-0.003	0.002
M3: Residual Invariance Model	17737.036*	5238	0.033	0.906	0.904	M3 vs. M2	253.802*	0	0.000	0.000	0.000
School status invariance (school ( n =3,984) vs. non-school ( n =510))
Community domain
M0: Configural Invariance Model (Baseline model)	6940.439*	946	0.053	0.895	0.882
M1: Metrics invariance Model	6794.036*	972	0.052	0.898	0.889	M1 vs. M0	62.405*	26	-0.001	0.003	0.007
M2: Scalar invariance Model	6929.500*	1048	0.050	0.897	0.896	M2 vs. M1	214.910*	76	-0.002	-0.001	0.007
M3: Residual Invariance Model	6929.500*	1048	0.050	0.897	0.896	M3 vs. M2	95.098*	0	0.000	0.000	0.000
Family domain
M0: Configural Invariance Model (Baseline model)	12416.379*	2088	0.047	0.873	0.863
M1: Metrics invariance Model	11286.542*	2127	0.044	0.887	0.880	M1 vs. M0	62.012	39	-0.003	0.014	0.017
M2: Scalar invariance Model	11438.288*	2265	0.042	0.887	0.888	M2 vs. M1	317.330*	138	-0.002	0.000	0.008
M3: Residual Invariance Model	11438.288*	2265	0.042	0.887	0.888	M3 vs. M2	106.261*	0	0.000	0.000	0.000
Peer/Individual domain
M0: Configural Invariance Model (Baseline model)	15118.341*	4948	0.030	0.911	0.903
M1: Metrics invariance Model	14346.315*	5004	0.029	0.918	0.912	M1 vs. M0	83.588	56	-0.001	0.007	0.009
M2: Scalar invariance Model	14630.839*	5238	0.028	0.918	0.915	M2 vs. M1	669.006*	234	-0.001	0.000	0.003
M3: Residual Invariance Model	14630.839*	5238	0.028	0.918	0.915	M3 vs. M2	201.294*	0	0.000	0.000	0.000

Note. All χ2 statistics were significant at p <. 001, RMSEA = Root Mean Square Error of Approximation; CFI = Comparative Fit Index; TLI = Tucker–Lewis Index; ∆ = differences between the comparison and nested model.

p < .001.

Regarding the results of the second set of analysis, it is suggested that all the risk and protective factor scales are valid for both school and non-school youth. Specifically, first, the fit indices for the baseline (configural) models are acceptable across four domains. This supports the configural invariance of all four domains in school and non-school youth subsamples. Second, given that configural invariance was supported, we examined metrics invariance, and the results of model comparison between configural invariance model and loading invariance model supported metric invariance (range of ∆RMSEA = .001 – .003, range of ∆CFI = .003 – .014, range of ∆TLI = .007 – .017), indicating that the item factor loadings are consistent for school and non-school youth subsamples. Third, given support for metric invariance, scalar invariance was tested and supported (range of ∆RMSEA = .001 - .002, range of ∆CFI = .000 – .001, range of ∆TLI = .003 – .008). This indicates that the measurement intercepts of each risk and protective scale are consistent across school and non-school youth. Last, given that scalar invariance was supported, we examined residual invariance. Changes in model fit indices indicate support for residual invariance (∆RMSEA = .000, ∆CFI = .000, ∆TLI = .000), suggesting that residuals of each risk and protective scale are invariant across school and non-school youth. Therefore, our results showed that each domain’s risk and protective factors are generalizable to males and females and youth in and out of school.

Discussion and Conclusion

Identifying the risk and protective factors that influence problem behaviors like violence, drug use, and gang involvement is critical for prevention and intervention programs. International development organizations are rapidly adopting and implementing risk and protective-oriented diagnostic instruments to target at-risk populations (Katz et al., 2022; Maguire et al., 2011). Although several studies have validated youth risk and protective factor scales in the United States, few have examined their validity in developing nations. Using a sample of Honduran youth, the present study examined the internal consistency, construct validity, concurrent validity, and measurement invariance across gender and school status for 38 risk and protective factors contained in the IMC.

Our findings showed that many IMC scales are reliable and valid for use in Honduras for their application to prevention/intervention programming. Specifically, for reliability, 34 of the 38 scales showed acceptable levels of internal consistency (ω ≥ .50), and 26 of the 38 scales demonstrated preferable levels of internal consistency (ω ≥ .70) (See Watkins, 2017). Four scales failed to produce an ω coefficient due to the small number of items in the scales. Hence, these findings support prior research suggesting that scales contained in CTCYS exhibit reasonable levels of internal consistency in nations that are uniquely different from the United States, where the instrument was first developed (e.g., Morojele et al., 2002). Translating the instrument’s language from English to Spanish was the only necessary adaptation.

Our results related to construct validity showed that the majority of items in IMC scales retained adequate factor loadings with good model fits. It is important to note that five of the 173 items exhibited low factor loadings, and several model adjustments were implemented by adding covariances between highly correlated items to improve construct validity of the IMC scales. These refinements were theoretically and substantively motivated, with correlated residuals specified only among items that exhibit clear semantic overlap, shared referents, or closely related behavioral content. Incorporating these parameters improved model fit while maintaining the conceptual coherence and construct validity of the IMC scales. Collectively, these findings underscore that risk and protective factor scales validated in the United States should not be assumed to function equivalently across contexts and warrant careful examination of their construct validity prior to adoption in new settings.

All 27 risk factor subscales and seven protective subscales demonstrated satisfactory concurrent validity, while four protective factor scales yielded mixed results. These findings provide support for the concurrent validity of the risk and protective factor scales across a wide variety of problem behaviors related to delinquency, gang involvement, drug use and drug sales, and gun carrying. It is consistent across the majority of the literature on CTCYS risk factors (e.g., Arthur et al., 2007; Hawkins et al., 2008). Conversely, our findings revealed weak concurrent validity for four protective subfactors. It is similar to those reported by Razali and Kliewer (2015), who also reported mixed support for the concurrent validity of the CTCYS protective factor scales among adolescents and young adults in Malaysia. Unexpected findings related to protective factors may occur because they are often conceptualized as moderators of risk factors and are less frequently examined for their direct effects (see Pardini et al., 2012). Taken together, at a minimum, our findings might suggest that some protective factors do not play the same role in protecting youth from problem behaviors examined in the present study.

Our findings on measurement invariance across gender and school status showed that the scales are generalizable across these subgroups. Risk and protective factors across each domain fit the data, and the factors within each domain were unique and possessed equality across factors within each domain. These findings suggest that different scales in IMC are not required when examining males and females and school and non-school-attending youth. They also indicate that in the future, at least in Honduras, the instrument and its adapted scales can be used to compare variation between subgroups without substantial concern about in-built measurement bias.

This paper makes a methodological contribution to the body of literature on youth risk and protective factors and their relationship with problem behavior in a nation with some of the highest rates of violence in the world. Conservatively, our findings suggest that the IMC can help understand causes of youth problem behavior and assist policymakers and practitioners in accurately identifying and supporting the most at-risk youth. In addition, the results provide a substantive contribution toward understanding risk and protective factors associated with youth problem behaviors. Our findings, for example, showed that peer-individual factors have a particularly robust relationship with several problem behaviors in Honduras. This can be most clearly seen when we convert our point biserial coefficients to Cohen’s d effect size $(d = \frac{2 r}{\sqrt{1 - r^{2}}}) .$ In doing so, our findings indicated that having delinquent peers strongly correlates with all problem behaviors, with effect sizes ranging from .54 for carrying a weapon to .80 for gang involvement. Similarly, modest to large effect sizes were observed for friends’ use of drugs, interaction with antisocial peers, and intention to use drugs.

Our findings, combined with prior research in developing and developed nations, suggest that peer-individual factors are vital to understanding delinquency. For example, Salas-Wright et al. (2013) reported that having delinquent peers, being in a gang, having low social support, and low levels of spirituality were significantly associated with violence and delinquency among Salvadorian school youth. Katz and Fox (2010) reported that in Trinidad, gang involvement was related to such peer individual factors as early initiation of anti-social behavior, intention to use drugs, having anti-social peers, having peers who use drugs, and low levels of belief in moral order. Similar findings have been reported in the United States (Hill et al., 1999), Australia (Beyers et al., 2004), and El Salvador (Webb et al., 2016).

Likewise, youth who have a family history of antisocial behavior are much more likely to be involved in all of the measured problem behaviors, varying from an effect size of .39 for carrying a weapon to about .72 for drug use. Youth who have more positive perceptions about the availability of drugs in their community are more likely to be involved in problem behaviors, varying from an effect size of .35 for drug selling to .70 for drug use. For youth who are attending school, low commitment to school is a significant risk factor for involvement in all measured problem behaviors, varying from an effect size of .28 for gang involvement, drug selling, and carrying a weapon to .50 for drug use. These findings also support prior studies of risk factors in developing and developed countries.

Directed prevention and intervention programming that addresses risk factors like those identified in the present study can potentially reduce not only risk factors but also a variety of problem behaviors. For example, some risk-based intervention programs have positively impacted alcohol use, antisocial-aggressive behavior, delinquency, criminal behavior, illicit drug use, tobacco, and violence (e.g., CTC; Multisystemic Therapy (MST)). With the above said, only a handful of risk and protective-based prevention and intervention programs have been evaluated in the United States (e.g., Feinberg et al., 2010), and only two evaluations have examined their impact in Central America and the Caribbean. Both of these studies, one conducted in Honduras, reported that programming significantly reduced risk factors but did not reduce delinquency (Katz et al., 2022; Stahlberg et al., 2022). Increased emphasis should be placed on evaluating the implementation of risk-based intervention programs in Honduras, El Salvador, and the Caribbean to test the effectiveness of this approach and determine whether reducing risk factors among youth impacts future delinquency.

The present study is not without its limitations. The study employed a cross-sectional design. Research in the future may examine the validity of IMC with a longitudinal design to determine whether risk and protective factors possessed at an earlier age result in future problem behavior. Since this study did not employ a random sample of participants, results might not be generalizable to the broader Honduran youth population. Future research may consider using a random sampling technique to ensure a greater generalizability of the findings. In addition, although administration procedures were designed to reduce age-related differences in comprehension, the broad age range of the sample may still reflect developmental variation in the manifestation of some constructs. Future research should more explicitly examine age-based measurement invariance and developmental differences in scale functioning. Finally, items in IMC are based on the respondent’s perceptions, not objective indicators of risk. For example, respondents’ perceptions of drug availability might be impacted by other contextual factors (e.g., social networks, parental beliefs) rather than the actual availability of firearms and drugs in their neighborhood. Further studies incorporating objective risk measures, like community surveys or environmental assessments, alongside respondents’ perceptions, may be needed to better understand the actual risks and how contextual factors influence these perceptions.

Despite these limitations, our findings contribute to the existing literature on the reliability and validity of risk and protective factors by examining the IMC in a sample of youth residing in some of the most dangerous communities in Honduras. Overall, our findings provide compelling evidence that the risk and protective factor scales within the IMC demonstrate good internal consistency, construct validity, concurrent validity, and measurement invariance. This tool may, therefore, inform the development of evidence-based crime prevention/intervention programs aimed at identifying risk and protective factors, ultimately contributing to the reduction of crime and delinquency among youth in Honduras.

Supplemental Material

sj-docx-1-cjb-10.1177_00938548261435853 – Supplemental material for At-risk Youth in Honduras: A Psychometric Evaluation of the Instrumento de Medicion de Comportamientos (IMC)

Supplemental material, sj-docx-1-cjb-10.1177_00938548261435853 for At-risk Youth in Honduras: A Psychometric Evaluation of the Instrumento de Medicion de Comportamientos (IMC) by Hyunjung Cheon, Charles M. Katz and Yi Zheng in Criminal Justice and Behavior

Footnotes

Authors’ Note:

The authors would like to thank Robyn Braverman and Guillermo Céspedes for their leadership and support throughout the project, Axel Rivera for all of his assistance on matters related to data collection and interpretation. The manuscript is the result of research conducted under Award No. AID 522- TO-16-00001, which was a contract from USAID to Creative Associates, for which the authors served as the evaluators.

ORCID iDs

Hyunjung Cheon

Charles M. Katz

Supplemental Material

Supplemental Appendices A to C and Supplemental Figures S1 to S4 are available in the online version of this article at .

Notes

Author Biographies

Hyunjung Cheon, PhD, is an assistant professor in the Department of Criminal Justice and Security Studies at the University of Texas at El Paso. Her research focuses on comparative criminology, race/ethnicity, policing, and quantitative methods.

Charles M. Katz, PhD, is a professor in the School of Criminology and Criminal Justice and Watts Family Director of the Center for Violence Prevention and Community Safety at Arizona State University. His research involves collaborating with agencies to increase their organizational capacity to identify and respond to crime affecting local communities.

Yi Zheng, PhD, is an associate professor in the Mary Lou Fulton College of Teaching and Learning Innovation and the School of Mathematical and Statistical Sciences at Arizona State University. Dr. Zheng is a psychometrician who studies how educational and psychological measurement instruments (e.g., tests, scales, assessments) are best designed and validated.

References

Akers

R. L.

(1973). Deviant behavior: A social learning approach. Wadsworth Pub Co.

Andrews

D. A.

Bonta

Wormith

J. S.

(2011). The risk-need-responsivity (RNR) model: Does adding the good lives model contribute to effective crime prevention? Criminal Justice and Behavior, 38(7), 735–755. https://doi.org/10.1177/0093854811406356

Arthur

M. W.

Briney

J. S.

Hawkins

J. D.

Abbott

R. D.

Brooke-Weiss

B. L.

Catalano

R. F.

(2007). Measuring risk and protection in communities using the Communities That Care Youth Survey. Evaluation and Program Planning, 30(2), 197–211. https://doi.org/10.1016/j.evalprogplan.2007.01.009

Arthur

M. W.

Hawkins

J. D.

Pollard

J. A.

Catalano

R. F.

Baglioni

A. J.

Jr . (2002). Measuring risk and protective factors for substance use, delinquency, and other adolescent problem behaviors: The Communities That Care Youth Survey. Evaluation Review, 26(6), 575–601. https://doi.org/10.1177/0193841X0202600601

Beyers

J. M.

Toumbourou

J. W.

Catalano

R. F.

Arthur

M. W.

Hawkins

J. D.

(2004). A cross-national comparison of risk and protective factors for adolescent substance use: The United States and Australia. Journal of Adolescent Health, 35(1), 3–16. https://doi.org/10.1016/j.jadohealth.2003.08.015

Blum

R. W.

Ireland

(2004). Reducing risk, increasing protective factors: Findings from the Caribbean youth health survey. Journal of Adolescent Health, 35(6), 493–500. https://doi.org/10.1016/j.jadohealth.2004.01.009

Catalano

R. F.

Hawkins

J. D.

(1996). The social development model: A theory of antisocial behavior. In Hawkins

J. D.

(Ed.), Delinquency and crime: Current theories (pp. 149–197). Cambridge University Press.

Chen

F. F.

(2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling: A Multidisciplinary Journal, 14(3), 464–504. https://doi.org/10.1080/10705510701301834

Cheung

G. W.

Rensvold

R. B.

(2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9(2), 233–255. https://doi.org/10.1207/S15328007SEM0902_5

10.

Cohen

Aiken

L. S.

West

S. G.

(1999). The problem of units and the circumstance for POMP. Multivariate Behavioral Research, 34(3), 315–346. https://doi.org/10.1207/S15327906MBR3403_2

11.

Cook

T. D.

Campbell

D. T.

(1979). Quasi-experimentation: Design & analysis issues for field settings. Houghton Mifflin.

12.

Cortina

J. M.

(1993). What is coefficient alpha? An examination of theory and applications. Journal of Applied Psychology, 78(1), 98–104. https://doi.org/10.1037/0021-9010.78.1.98

13.

CRAFFT. (n.d.). CRAFFT substance use screening instrument. https://www.crafft.org

14.

De Jesus

Hernandes

. (2019). Generalized violence as a threat to health and well-being: A qualitative study of youth living in urban settings in central America’s “Northern triangle.” International Journal of Environmental Research and Public Health, 16(18), 3465. https://doi.org/10.3390/ijerph16183465

15.

Desa

Van de Vijver

F. J.

Carstens

Schulz

(2019). Measurement invariance in international large-scale assessments: Integrating theory and method. In Johnson

T. P.

Pennell

B.-E.

Stoop

Dorer

(Eds.), Advances in comparative survey methodology (pp. 881–910). Wiley.

16.

Fairnington

(2004). Communities that care: A case study of regeneration from Wales. Critical Public Health, 14(1), 27–36. https://doi.org/10.1080/09581590310001647489

17.

Farrington

D. P.

Loeber

Ttofi

M. M

. (2012). Risk and protective factors for offending. In Welsh

B. C.

Farrington

D. P.

(Eds.), The Oxford handbook of crime prevention (pp. 46–69). Oxford University Press.

18.

Feinberg

M. E.

Jones

Greenberg

M. T.

Osgood

D. W.

Bontempo

(2010). Effects of the communities that care model in Pennsylvania on youth risk and problem behaviors. Prevention Science, 11(2), 163–171. https://doi.org/10.1007/s11121-009-0161-x

19.

Felitti

V. J.

Anda

R. F.

Nordenberg

Williamson

D. F.

Spitz

A. M.

Edwards

Marks

J. S.

(1998). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: The Adverse Childhood Experiences (ACE) Study. American Journal of Preventive Medicine, 14(4), 245–258. https://doi.org/10.1016/S0749-3797(98)00017-8

20.

Flynn

R. J.

(2008). Communities that care: A comprehensive system for youth prevention and promotion, and Canadian applications to date. IPC Review, 2(3), 83–106.

21.

Frank-Vitale

Hoffnung-Garskof

(2025). Honduras, gangs, and asylum law. Latin American Perspectives, 52(2), 105–124. https://doi.org/10.1177/0094582X251321833

22.

Glaser

R. R.

Horn

M. L. V.

Arthur

M. W.

Hawkins

J. D.

Catalano

R. F.

(2005). Measurement properties of the Communities that Care® Youth Survey across demographic groups. Journal of Quantitative Criminology, 21, 73–102. https://doi.org/10.1007/s10940-004-1788-1

23.

Haggerty

R. J.

Mrazek

P. J.

(Eds.). (1994). Reducing risks for mental disorders: Frontiers for preventive intervention research. National Academy Press.

24.

Hawkins

J. D.

Brown

E. C.

Oesterle

Arthur

M. W.

Abbott

R. D.

Catalano

R. F.

(2008). Early effects of Communities That Care on targeted risks and initiation of delinquent behavior and substance use. Journal of Adolescent Health, 43(1), 15–22. https://doi.org/10.1016/j.jadohealth.2008.01.022

25.

Hawkins

J. D.

Catalano

R. F.

Arthur

M. W.

(2017). Promoting science-based prevention in communities. In Natarajan

(Ed.), Drug Abuse: Prevention and Treatment (pp. 121–146). Routledge.

26.

Hawkins

J. D.

Catalano

R. F.

Miller

J. Y.

(1992). Risk and protective factors for alcohol and other drug problems in adolescence and early adulthood: Implications for substance abuse prevention. Psychological Bulletin, 112(1), 64–105. https://doi.org/10.1037/0033-2909.112.1.64

27.

Hill

K. G.

Howell

J. C.

Hawkins

J. D.

Battin-Pearson

S. R.

(1999). Childhood risk factors for adolescent gang membership: Results from the Seattle Social Development Project. Journal of Research in Crime and Delinquency, 36(3), 300–322. https://doi.org/10.1177/0022427899036003003

28.

Hirschi

(1969). Causes of delinquency. University of California Press.

29.

Howell

J. C.

Egley

Jr. (2005). Moving risk factors into developmental theories of gang membership. Youth Violence and Juvenile Justice, 3(4), 334–354. https://doi.org/10.1177/1541204005278679

30.

Bentler

P. M.

(1999). Cutoff criteria for fit indices in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. https://doi.org/10.1080/10705519909540118

31.

Insight Crime. (2024). InSight crime’s 2023 homicide round-up. https://insightcrime.org/wp-content/uploads/2023/08/InSight-Crimes-2023-Homicide-Round-Up-Feb-2024-2.pdf

32.

Jonkman

Aussems

Steketee

Boutellier

Cuijpers

(2015). Prevention of problem behaviours among adolescents: The impact of the communities that care strategy in the Netherlands (2008–2011). International Journal of Developmental Science, 9(1), 37–52. https://doi.org/10.3233/DEV-13121

33.

Katz

C. M.

Cheon

Hedberg

E. C.

Decker

S. H.

(2022). Impact of family-based secondary prevention programming on risk, resilience, and delinquency: A 6-month follow up within a randomized control trial in Honduras. Justice Quarterly, 39(6), 1237–1262. https://doi.org/10.1080/07418825.2021.1967425

34.

Katz

C. M.

Fox

A. M.

(2010). Risk and protective factors associated with gang-involved youth in Trinidad and Tobago. Revista panamericana de salud publica = Pan American journal of public health. 27, 187–202.

35.

Koonings

Kruijt

(Eds.). (2007). Fractured cities: Social exclusion, urban violence and contested spaces in Latin America. Zed Books.

36.

Landa-Blanco

Cheon

Flores

L. G. R.

Spohn

Katz

C. M.

(2020). Violence in Honduras from 2008 to 2018. Injury Prevention: Journal of the International Society for Child and Adolescent Injury Prevention, 26(2), 191–193. https://doi.org/10.1136/injuryprev-2019-043536

37.

Loeber

Farrington

D. P

. (2000). Young children who commit crime: Epidemiology, developmental origins, risk factors, early interventions, and policy implications. Development and Psychopathology, 12(4), 737–762. https://doi.org/10.1017/S0954579400004107

38.

Maguire

E. R.

Wells

Katz

C. M.

(2011). Measuring community risk and protective factors for adolescent problem behaviors: Evidence from a developing nation. Journal of Research in Crime and Delinquency, 48(4), 594–620. https://doi.org/10.1177/0022427810395148

39.

Masten

A. S.

Coatsworth

J. D

. (1998). The development of competence in favorable and unfavorable environments: Lessons from research on successful children. American Psychologist, 53(2), 205–220.

40.

McDonald

R. P.

(1999). Test theory: A unified treatment. Erlbaum.

41.

McNeish

(2018). Thanks coefficient alpha, we’ll take it from here. Psychological Methods, 23(3), 412–433. https://doi.org/10.1037/met0000144

42.

Meade

A. W.

Johnson

E. C.

Braddy

P. W.

(2008). Power and sensitivity of alternative fit indices in tests of measurement invariance. Journal of Applied Psychology, 93(3), 568–592. https://doi.org/10.1037/0021-9010.93.3.568

43.

Morojele

N. K.

Flisher

A. J.

Muller

Ziervogel

C. F.

Reddy

Lombard

C. J.

(2002). Measurement of risk and protective factors for drug use and anti-social behavior among high school students in South Africa. Journal of Drug Education, 32(1), 25–39. https://doi.org/10.2190/MJDD-PC1G-4KUH-C1YW

44.

Murray

A. L.

Eisner

Ribeaud

(2018). Development and validation of a brief measure of violent thoughts: The Violent Ideations Scale (VIS). Assessment, 25(7), 942–955. https://doi.org/10.1177/1073191116667213

45.

Ohene

S. A.

Ireland

Blum

R. W.

(2005). The clustering of risk behaviors among Caribbean youth. Maternal and Child Health Journal, 9, 91–100. https://doi.org/10.1007/s10995-005-2452-6

46.

Pardini

D. A.

Loeber

Farrington

D. P.

Stouthamer–Loeber

(2012). Identifying direct protective factors for nonviolence. American Journal of Preventive Medicine, 43(2), S28–S40. https://doi.org/10.1016/j.amepre.2012.04.024

47.

Razali

M. M.

Kliewer

(2015). Validation of the communities that care measure adapted for use in Malaysia. International Perspectives in Psychology, 4(4), 267–280. https://doi.org/10.1037/ipp0000039

48.

Roosa

M. W.

Zeiders

K. H.

Knight

G. P.

Gonzales

N. A.

Tein

J.-Y.

Saenz

O’Donnell

Berkel

(2011). A test of the social development model during the transition to junior high with Mexican American adolescents. Developmental Psychology, 47(2), 527–537. https://doi.org/10.1037/a0021269

49.

Salas-Wright

C. P.

Olate

Vaughn

M. G.

(2013). Assessing empathy in Salvadoran high-risk and gang-involved adolescents and young adults: A Spanish validation of the Basic Empathy Scale. International Journal of Offender Therapy and Comparative Criminology, 57(11), 1393–1416. https://doi.org/10.1177/0306624X12455170

50.

Shaffer

Yates

T. M.

(2010). Identifying and understanding risk factors and protective factors in clinical practice. In Compton

M. T.

(Ed.), Clinical Manual of Prevention in Mental Health (pp. 29–48). American Psychiatric Publishing, Inc.

51.

Stahlberg

S. G.

Díaz-Cayeros

Pizatella-Haswell

(2022). Supporting youth and families to prevent risky youth behavior and delinquency: An impact evaluation of a family counseling program in the Caribbean. Children and Youth Services Review, 142, 106645. https://doi.org/10.1016/j.childyouth.2022.106645

52.

Sutherland

E. H

. (1973). A statement of the theory. In Schuessler

K. F.

(Ed.), On analyzing crime (pp. 7–12). University of Chicago Press.

53.

Tabachnick

B. G.

Fidell

L. S.

Ullman

J. B.

(2007). Using multivariate statistics (Vol. 5). Pearson.

54.

Thapa

Cohen

Guffey

Higgins-D’Alessandro

(2013). A review of school climate research. Review of Educational Research, 83(3), 357–385. https://doi.org/10.3102/0034654313483907

55.

Thurow

C. F.

Nunes

Schneider

D. R.

(2021). Psychometrics and cultural adaptations of the Community That Care Youth Survey: A systematic review. Journal of Community Psychology, 49(7), 2892–2904. https://doi.org/10.1002/jcop.22529

56.

United Nations Office on Drugs and Crime. (2023). Global study on Homicide 2023. UNODC Research. https://www.unodc.org/documents/data-and-analysis/gsh/2023/Global_study_on_homicide_2023_web.pdf

57.

Vandenberg

R. J.

Lance

C. E.

(2000). A review and synthesis of the measurement invariance literature: Suggestions, practices, and recommendations for organizational research. Organizational Research Methods, 3(1), 4–69. https://doi.org/10.1177/109442810031002

58.

Vincent

M. G.

Skeem

Weber

(2025, April). Youth protective factors study: Supervision based on risks, strengths, and development. https://csgjusticecenter.org/wp-content/uploads/2025/03/NIJ-Protective-Factors-Slides-4.23.25-FINAL.pdf

59.

Watkins

M. W.

(2017). The reliability of multidimensional neuropsychological measures: From alpha to omega. The Clinical Neuropsychologist, 31(6–7), 1113–1126. https://doi.org/10.1080/13854046.2017.1317364

60.

Webb

V. J.

Nuño

L. E.

Katz

C. M.

(2016). Influence of risk and protective factors on school-aged youth involvement with gangs, guns, crime and delinquency: Findings from the El Salvador Youth Survey. Center for Violence Prevention and Community Safety, Arizona State University.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.39 MB

0.00 MB