|Home | About | Journals | Submit | Contact Us | Français|
The purpose of this study was to investigate the psychometric properties of a child self efficacy scale for learning, peer interactions, and resisting pressure to use drugs, to use in an elementary school drug prevention education program based on social cognitive theory. A diverse cohort of 392 4th and 5th grade students completed the 20-item self efficacy scale and social support and social skills instruments. The results provide evidence for a valid and reliable 3-factor self efficacy scale. Subscale internal consistency reliability was good to excellent (Cronbach’s alpha = 0.75, 0.83, 0.91). Construct validity was supported by correlations between each subscale and social skills, social support, and demographic data. The scale has potential as a tool to measure self efficacy in children related to learning, peer interactions, and resisting peer pressure to use drags and to help shape drag education programs.
Social Cognitive Theory (SCT) provides a comprehensive and well-supported conceptual framework for understanding the factors that influence human behavior and the processes through which learning occurs, offering insight into a wide variety of health related issues. SCT explains human functioning in terms of a 3-part interactive model in which behavior, cognitive and other personal factors, and environmental events all operate as interacting determinants of each other. In other words, performing a behavior is a dynamic action that depends on aspects of the person, the environment, and the behavior itself. Each of these factors (behavior, person, and environment) influences the others simultaneously and the factors are constantly interacting (Bandura, 1986). SCT is attractive to apply to drug education because it not only illuminates the dynamics of individual behavior but also gives direction to the design of intervention strategies to influence behavior change (McGahee, Kemp, & Tingen, 2000). The use of SCT in the development of drag education programs is supported by it application in primary prevention programs focused on health concerns. Examples include family planning decision making (Ha, Jayasuriya, & Owen, 2005), a HIV prevention program in adolescents (Dilorio, Resnicow, McCarty, De, Dudley, Wang, ct al., 2006), a nutrition education program (Powers, Struempler, Guarino, & Parmer, 2005), and a poison prevention education program (Schwartz, Howland, Mercurio-Zappala, & Hoffman, 2003). An example of an existing drug education program using SCT is Shegog and others’ (2005) middle school program to help students cope with pressure to use tobacco products.
Social cognitive theory implicates self efficacy, the belief in one’s ability to perform a certain task, as a pivotal construct in understanding and modifying human behavior. According to Bandura (1999), self efficacy is a central mechanism in the exercise of agency. An individual’s belief about her or his capacity to perform a task or withstand difficulty will greatly determine how they actually behave. Individuals who doubt their ability in a given domain are more likely to avoid difficult tasks, set low aspirations, and make minimal commitment to goals they set in that area. Persons with strong self efficacy expectations in a particular domain, however, approach tasks as challenges rather than threats, persist when their efforts fail, feel in control of potential stressors, and maintain commitments to goals. Furthermore, a secure sense of self efficacy fosters positive social relationships, whereas insecure self efficacy tends to breed socially alienating behaviors (Bandura, Barbaranelli, Caprara, & Pastorelli, 1996). Evidence for the predictive ability of self efficacy in children has been demonstrated in cognitive tasks, such as mathematics achievement and other learning activities (Zimmerman & Cleary, 2006), in social adaptation such as childhood resiliency (Cowen, Work, High-tower, Wyman, Parker, & Lotyczewski, 1991), physical activity (Kane, 2007), food related behaviors (Parcel et al., 1995; Perry, Ayala, Lebow, & Hayden, 2007), depression (Bandura et al, 1996), and anxiety disorders (Muris, 2002). Bandura, Pastorelli, Barbaranelli, and Caprara (1999) found the predictiveness of what they called perceived self-regulatory self efficacy to resist pressure to engage in high risk activities (i.e., drug use, early sexual activity). And they highlight the pressure on youth to engage in high risk activities and centrality of young people’s perceived self-regulatory self efficacy to their health.
Bandura (1999) discussed that the establishment and maintenance of social relationships serve as buffers against stressors that can lead to negative health outcomes (i.e., depression). However, these constructs are not self-forming entities waiting passively to buffer children and teenagers against stressors; children and adolescents must find, create, and maintain supportive relationships for themselves (Bandura et al., 1996). In this way, self efficacy is related to both social support and social skills. Researchers (Demaray & Malecki, 2002; Haggerty, Sherrod, Garmezy, & Rutter, 1996) have found that social support and social skills reduce vulnerability to stress, depression, physical illness, and poor health choices such as cigarette, alcohol, and marijuana use. Researchers have shown that early internalizing behaviors are predictive of later externalizing behaviors and that external behaviors are frequently comorbid with other problems, such as depression and hyperactivity (Demaray & Malecki, 2003; Haggerty et al., 1996).
Presumably self efficacy helps shape qualities and/or self-views in the growing children that have important adaptive value. With increasing pressure to use and availability of tobacco, alcohol, and marijuana at an earlier age, children’s belief that he or she can successfully execute the behavior to resist drug use needs to be better understood (Annie E. Casey Foundation, 2007; Blanchard, Gurka, & Black-man, 2006). Self regulation skills related to learning and managing peer relations have been shown to be important for controlling and changing behavior (Locke & Latham, 2002). These social skills as well as social support are integral to the motivation, attentiveness, and behavior attainment of young people to make health behavior choices such as not to use drugs (Cullen, Baranowski, & Smith, 2001).
In recently published studies, interventions have targeted changes in self efficacy related to a number of health concerns among children including fat and sugar intake (Rinderknecht & Smith, 2004), fruit and vegetable consumption (Molaison, Connell, Stuff, Yadrick, & Bogle, 2005), nutrition education (Leach & Yates, 2008; Powers et al., 2005), and children and adolescents seizure disorder management (Caplin, Austin, Dunn, Shen, & Perkins, 2002). School-based drug education programs have shown promise to increase adolescents’ self efficacy for coping with pressure to smoke (Shegog et al., 2005). Building on these studies and using the theoretical underpinnings of SCT, an elementary school drug education program was developed, EYSP–Elementary Youth Support Program (Becker, Kuntz, & Fertman, 2008; Council on Drug & Alcohol Abuse, 2006). At the center of the program is the use of four strategies to build self efficacy (Bandura, 2004): mastery experience, verbal persuasion and reassurance, social modeling, and reducing stress. Mastery activities focus on breaking down learning and social skills into practical and doable steps. Furthermore, since self efficacy is a behavior specific construct, the mastery activities are explicit in their focus on learning skills and strategies not to use tobacco, alcohol, or marijuana. Verbal persuasion and reassurance strategies include the teachers leading the program identifying where students have been successful in changing a negative behavior or in acquiring a positive behavior and stating how students can do the same in learning, peer relations, and not using drugs. Positive peer models, kinship networks, and family members are used to show students that others like themselves can be successful in school, have friends, and not use drugs. Finally, program activities focus on lowering the children’s stress levels through both personal strategies to reduce anxiety in relation to school, peers, and drugs (tobacco, alcohol, and marijuana), and school and community efforts aimed at policies and access to supportive services (e.g., health services, after-school activities, family support). Finally, EYSP recognizes the importance of social skills and social support in drug education and attempts to build on this foundation by the use of the SCT framework to focus activities on not using tobacco, alcohol, or marijuana.
Given the central role of self efficacy in the theoretical underpinning of EYSP and children’s behavior not to use drugs, a reliable and valid measurement of self efficacy is crucial to a thorough understanding of the behavior change process and prevention of drug problems. Furthermore, critical to development of evidence-based drug education programs is that use of SCT as a theoretical base is valid and reliable instrumentation. Therefore, the current study was to investigate the psychometric properties of child self efficacy scale to use in EYSP as well as serve as a model for elementary drag prevention education program based on SCT. And in particular, drug education programs that utilize SCT strategies to increase self efficacy: mastery experience, verbal persuasion and reassurance, social modeling, and reducing stress.
Bandura pioneered work in the measurement of children’s self efficacy via the establishment of scales to measure children’s self efficacy on several dimensions (Bandura et al., 1999). Utilizing his social cognitive theory as a model, he proposed 28 items related to five constructs of self efficacy: academic achievement; learning; leisure and extracurricular activities; self-regulatory to resist pressure to engage in high-risk activities involving alcohol, drugs, and transgressive behavior; and social relations. He then administered the survey to 282 5th and 6th grade students in a village near Rome, Italy. His research verified the influential role played by beliefs of personal efficacy in childhood development (Bandura et al., 1999). In later work he presents a larger scale focused on nine domains but without any reported studies related to the instrument’s psychometrics (Bandura, 2006).
Review of Bandura’s early work in self efficacy measurement although promising does reveal gaps in the literature. First, Bandura’s sample outside Rome was largely homogeneous in terms of race, ethnicity, and setting, since all inhabitants were rural. Second, it would be instructive to test these scale items among even younger students, since in the United States risk taking behavior can start extremely early (Annie E. Casey Foundation, 2007; Blanchard et al., 2006) and since children encounter different stressors, have different skills, different social relationships, and present different emotional/behavior problems in elementary as compared to middle and high school (Mash & Barkley, 1996). Finally, Bandura’s attention to social support and social skills highlights that it may be instructive to investigate these constructs in relationship to self efficacy.
Based on Bandura’s early work, other investigators have pursued the development of children’s self efficacy measures. For example, Cowen, Work, High-tower, Wyman, Parker, and Lotyczewski (1991) did preliminary studies of a childhood self efficacy scale resulting in a 20-item provisional scale that did necessarily reflect all aspects of a generalized self efficacy construct. This work calls attention to self efficacy as a behavior-specific construct and the need to view self efficacy within the context of a specific behavior domain. Other early work in the development of self efficacy measures for youth include Parcel and others’ (1995) elementary student self efficacy scale for diet related behaviors. A measure of high school students’ drug resistance self efficacy was developed by Jenkins, Nolan, and Rieder (1997). More recently investigators have focused on children’s self efficacy related to seizure management (Caplin et al., 2002), depression (Tonge, King, Klimkeit, Melvin, & Gordon, 2005), chronic pain assessment (Bursch, Tsao, Meldrum, & Zeltzer, 2006), physical activity (Kane, 2007), and physical activity and healthy food choices (Perry et al., 2007).
The purpose of the current study was to develop and validate a scale measuring self efficacy related to learning, peer interactions and resisting pressure to use drugs in an ethnically and geographically diverse cohort of 4th and 5th grade children. We hypothesized that factor analysis would reveal face-valid subscales related to self efficacy related to learning, peer relations and resisting drug use with strong internal consistency reliability. Additionally, we hypothesized that our measures would correlate with established measures of social support and social skills.
We undertook the development of this scale as part of program evaluation for an elementary school social skills and social support curriculum sponsored by a state Department of Education.
The participants were all 4th and 5th grade students from four elementary schools: one suburban; one urban; and two rural. The building demographics are summarized in Table 1.
Children were administered the evaluation measures in their classrooms by school staff as part of an educational curriculum evaluation. Via an honest broker system, evaluators were blinded to all unique student identifiers. Grade level, gender, and ethnicity were the only demographics provided for purpose of the program evaluation. The University of Pittsburgh Institutional Review Board approved this study.
Candidate self efficacy items were taken from Bandura’s scale (Bandura, 2006). Although the complete scale consists of 55 items representing nine domains of functioning, all domains were not relevant to the current investigation. Since our interest was in the area of social skills and social support as related to health choices, we eliminated items related solely to academic subjects (such as having confidence to learn algebra or science) or extracurricular activities (such as having confidence to play sports). We also eliminated an item dealing with self efficacy regarding sexual intercourse because of the developmental stage of our sample. We also eliminated several items that were confusing or ambiguous in pilot studies. To facilitate students’ understanding of the items, we changed the items’ tense from third to first person and slightly modified language on some items. Finally, the response set was changed from the child assigning a probability percentage (i.e., 20%) to a pictorial 5-point Likert scale (Kane, 2007). The 20 resultant items comprehensively represented the construct of self-efficacy related to learning, peer interactions, and drug use for elementary children.
Social support was measured with the Child and Adolescent Social Support Scale (CASSS; Malecki & Demaray, 2002) which assesses a child’s social support from four sources (12 items each): my teacher(s); my classmate(s); my parent(s); and my close friend. Since the academic focus of the teacher subscale was beyond the scope of the current research, we eliminated this subscale, reducing the number of items from 48 to 36. On the CASSS, children rate their prosocial perceptions based on a 6-point scale classifying their behaviors (never, almost never, some of the time, most of the time, almost always, and always). The CASSS has been well studied, and its psychometric properties are strong (Demaray & Malecki, 2002, 2003; Malecki & Demaray, 2002, 2006).
Social skills were measured with the Social Skills Rating System (SSRS; Gresham & Elliott, 1990) which assesses a child’s social skills on four subscales: cooperation; assertion; empathy; and self-control. The scale contains 34 items. The SSRS is backed by extensive research and was standardized on a large national sample of youth (Benes, 1995). It was the first scale to rate social skills providing separate norms for boys and girls ages 3–18 and for elementary students with and without disabilities (Gresham & Elliott, 1990).
School, grade level, gender, and ethnicity (Hispanic, Black, not Hispanic, White, not Hispanic, Asian/Pacific Islander, and American Indian/Alaskan Native) were recorded.
We first conducted principal component analysis with varimax orthogonal rotation on the 20-item perceived self efficacy items in order to determine the factor structure underlying the data. This analytic technique revealed a 3-factor structure (Table 2). We labeled the first factor “Perceived Learning Self Efficacy” (PLSE), since items with high loading on this factor measured perceived capability to manage one’s own learning and fulfill expectations (N = 11). We labeled the second factor “Perceived Peer Self Efficacy (PPSE), since items loading on this factor included perceived capability for peer relationships and for self assertiveness (N= 6). The third factor, “Perceived Drug Use Self Efficacy” (PDUSE) included items measuring perceived capability to resist pressure to engage in high risk activities (N = 3). Students responded to each item using a 5-point Likert scale. All items and factor loadings are listed in Table 2.
Second, we used Cronbach’s alpha to assess the internal consistency reliability of the resultant subscales (Table 2). We then used statistical testing to determine if subscale scores were different across various demographic characteristics (school, gender, grade, and race). For subscales 1 and 2 (PLSE and PPSE), we treated scale values as continuous variables (range = 1 to 5). Based on the histogram for sub-scale 3 (PDUSE), however, we dichotomized results on this subscale into two values: not at risk for drug use (raw PDUSE = 5, thus strongly agreed with each statement) or at risk for drug use (raw PDUSE < 5, thus did not strongly agree with each statement). This method of dichotomizing such scales has precedent; research has shown that young people who indicate any response lacking strong commitment to avoid substance use are substantially more susceptible to later substance use than those who show consistent, strong commitment to avoid substance use (Pierce, Choi, Gilpin, Farkas, & Merritt, 1996). For analyses involving sub-scales 1 and 2 with continuous outcomes, we used analysis of variance (ANOVA), and for analyses involving subscale 3 with a dichotomous outcome we used chi-squared analyses. For all analyses, we considered statistical significance P<0.05.
Finally, to further investigate the construct validity of the subscales, we examined the correlations between:
For these analyses, we used Pearson’s r, conservatively corrected with the Dunn-Sidak multiple comparison method (Abdi, 2007). Again statistical significance was considered P < 0.05.
The participants in this study included 392 children in grades 4 and 5 (Table 3). Participants were half female and one-third non-White. Suburban, urban, and rural populations were equally represented.
Table 4 presents the self efficacy and drug risk scores by demographic characteristics. Each of the subscales had good-to-excellent internal consistency reliability (Cronbach’s alpha = 0.83, 0.75, 0.91 for PLSE, PPSE, and PDUSE, respectively). Scores on the PLSE and PPSE did not differ by school, gender, grade level, or race. However, scores on the PDUSE differed by school, grade level, and race. Those in the suburban school were much less likely to be at risk for drug use than those in the urban and rural schools (3.0% vs. 29.4%, 14.3%, and 37.5%o respectively, P < 0.001). Compared with those in grade 5, those in grade 4 were more likely to be at risk for drug use (22.2%) vs. 13.4%, P < 0.05), and non-Whites were more likely than Whites to be at risk (28.9% vs. 12.5%), P< 0.001).
Table 5 presents correlations between self efficacy scores and social support and social skills scores. PLSE scores were highly correlated with each of the social support and social skills subscales (all P < 0.001). PPSE scores were highly correlated with each of the social support and social skills subscales (all P < 0.001). PDUSE scores were significantly associated with social support from classmates as well as social skills subscales related to cooperation, assertion, empathy, and total social skills.
In this study, we developed a 3-factor self efficacy scale for school aged children with strong content validity based on social cognitive theory. Each sub-scale had good-to-excellent internal consistency reliability, and subscale scores were related in ways we would expect to measures of demographics, social support, and social skills, lending a measure of construct validity to the scale.
Subscales included Perceived Learning Self Efficacy (PLSE, N = 11), Perceived Peer Self Efficacy (PPSE, N = 6), and Perceived Drug Use Self Efficacy (PDUSE, N = 3). The predictive validity of each of these elements of self efficacy is supported by findings of prior research (Bandura, 2006; Bandura et al., 1996; Zimmerman, Bandura, & Martinez-Pons, 1992).
As we expected, each subscale was at least somewhat positively skewed, since students at this developmental level tend to agree with scale items. This was particularly dramatic for the PDUSE. If it had been used as a continuous variable, the PDUSE would have exhibited a strong ceiling effect. However, the subscale represents useful information when converted to a dichotomous variable distinguishing those who strongly agree with all statements with all respondents. In this case, about 60% of students strongly agreed with all PDUSE statements; thus, this measure divides the sample roughly evenly into those at higher risk and those at lower risk for later substance use. There is strong precedent for this type of measure. In adolescents, for example, it is standard to measure susceptibility to smoking with Pierce’s valid and reliable 3-item scale: “Do you expect to smoke a cigarette sometime in the next year?”; “If your best friend gave you a cigarette, would you smoke it?”; “Do you expect to smoke a cigarette soon?” (Pierce et al., 1996). Although students respond to each of these items using a 4-point Likert scale, the results are dichotomized into those who are not susceptible (those who answer “definitely no” to all items) and those who are susceptible (everyone else). In a similar way, those who respond with any equivocal response to the PDUSE may be susceptible to future drug use. It will be instructive to test the predictive validity of this construct by longitudinally following students’ ultimate uptake of drag use.
We found that PLSE and PPSE scores were not associated with demographic characteristics. PDUSE scores, however, were significantly different according to school and race. This is consistent with prior data showing that drug use is associated with certain environmental and racial characteristics (Office of Applied Studies, 2006). Furthermore, this suggests that these differences in propensity for later drug use may be measurable as early as 4th and 5th grade. If the PDUSE is found to have predictive validity, it could be used to identify those at risk with sufficient time to intervene.
Although PDUSE scores were also associated with grade level, they were higher among 4th graders than 5th graders, which was an unexpected result. This could indicate a weakness in the measurement and a threat to its validity. However, another potential explanation is that many curricular and extracurricular “anti-drug” programs begin in the 5th grade (Cain, Dickinson, Fernald, Bublitz, Dickinson, & West, 2006). Thus, it is possible that 5th graders may have greater anti-drug self efficacy because of timely exposure to these programs.
It was hypothesized that self efficacy would be correlated with both children’s social support and social skills. PLSE and PPSE scores were highly correlated with each of the social support and social skills subscales. However, drug use self efficacy was correlated with some of the subscales and not others. This may be because risk for drug use is moderated with other demographic or environmental variables (Office of Applied Studies, 2006). It is interesting to note that PDUSE was more strongly correlated with social support from classmates and friends than from parents. This is an interesting finding, since it suggests that lack of peer support may be even more important than lack of parental support in the decision to use substances. This is different than previous data suggesting that parental support is the most important type of social support related to substance use (Knight, Broome, Cross, & Simpson, 1998; Kumpfer & Alvarado, 2003; Sullivan, King, & Farrell, 2004).
It should be noted that development of a valid scale is a continuous, iterative process. In this study, we were able to determine some measure of content validity since we used an established model for item development. Furthermore, we obtained promising results related to construct validity since our scale values were correlated with expected constructs. However, this study was limited in that it did not address criterion validity of the scales. Further research will be necessary to determine whether young people scoring high on the PDUSE scale, for instance, do indeed go on to experiment frequently with substance use. Another potential limitation of these scales is the ceiling effect. However, this is a common limitation of survey measures such as this one, and we responded appropriately by defining data cut-off points with strong precedent and face validity. Finally, the findings identify associations between a child’s self efficacy, social support, and social skills but this study did not shed light on how such associations come about.
In conclusion, this study finds that self efficacy can be reliably measured among 392 4th and 5th graders. The 3-factor, 20-item scale developed here has potential to be used in drug education programs to measure self efficacy in children related to learning, peer relations, and resisting drug use. The instrument can be used by practitioners to help identify which students might need or are likely to benefit from self efficacy enhancing activities. For practitioners working in drug education programs based on social cognitive theory, the scale can be used to provide programmers with further insight into how elementary aged children judge their ability to not use tobacco, alcohol, and marijuana. Furthermore, a scan of the model programs listed on the National Registry of Evidence-based Programs and Practices (NREPP, 2008) reveals that while there are 23 model programs that target elementary school-aged populations and address alcohol, tobacco, or illegal drugs, SCT is not a predominate theoretical program underpinning. The instrument’s development and validation is a step toward wider use of SCT in drug education programs for elementary school students. Finally, further research of SCT-based drug education programs for elementary school students offers researchers links to an existing body of knowledge on a range of health behaviors and concerns that can enhance and support new drug education program development and implementation.