Measuring resilience among Dutch adolescents, a validation study of the Child and Youth Resilience Measure-12 (CYRM-12)
Master thesis Forensic Child and Youth Care Sciences Graduate School of Child Development and Education University of Amsterdam
H.E. Elzing, 12459437 Guidance: Measuring resilience among Dutch adolescents, a validation study of the Child and Youth Resilience Measure-12 (CYRM-12), G.J.J. Stams and L. V. van Dam Amsterdam, (March, 2020)
2 Abstract
This study focused on the validity of the Child Youth and Resilience Measure (CYRM-12). Internal structure, reliability, concurrent validity and measurement invariance for gender and education level were examined. The CYRM-12 was administered in a mixed gender sample (N = 763) of primarily Dutch (88.2%) and Surinamese (10.4%) adolescents from the general population. A one-factor structure with 10 items best fitted the data in a Confirmatory Factor Analysis (CFA). The reliability was satisfactory (α = .770). The high negative association between resilience (CYRM-12) and psychopathology (SCL-90-R) was considered indicative of concurrent validity. For gender strict invariance was established, and metric invariance for education level. No significant differences in resilience were found between male and female adolescents. Further research should establish measurement invariance of the Dutch 10-item version of the CYRM for migration background and clinical groups. Next, convergent, predictive and discriminant validity should be examined as well as test-retest reliability.
3 MEASURING RESILIENCE AMONG DUTCH ADOLESCENTTS, A VALIDATION
STUDY OF THE CYRM-12 Introduction
Adolescence can be described as a lifetime full of changes, in social, psychical and cognitive areas as well as significant life events, such as switching schools and changes in family structure (Dumont & Provost, 1999). However, some adolescents do not experience problems from these changes, because they are resilient. Resilience is thought to be important during adolescence, because it has the potential to protect against several risk factors. The Child and Youth Resilience Measure (CYRM-12) is a questionnaire to assess resilience in both children and adolescents. Several validation studies have been conducted on the CYRM-12, but not in the Netherlands (Van Dam et al., 2019). In addition, measurement invariance across gender and education has not yet been examined. Therefore, the present study focuses on the validity of the CYRM-12. It examines the internal structure of the CYRM-12, its reliability and concurrent validity by examining the association between resilience and (symptoms of) psychopathology.
The concept of resilience
Resilience is assumed to play an important role in the successful societal adaptation and healthy development of children and adolescents (Ungar, 2011). Notably, the psychological outcome after having experienced exposure to developmental or environmental risks depends on both individual and contextual characteristics (Strolin, Goltzman, Woodhouse, Suter, & Werrbach, 2016). Bronfenbrenner’s ecological system theory confirms that the environment (maturing biology, family, community and societal landscape) directly affects the development of a child (Ryan, 2001). Thus, the resilience of a child is influenced by its ecological system.
Resilience is a dynamic response to a multitude of biological, social, psychological and other environmental influences (Fraser, Galinsky, & Richman, 1999). Ungar and Liebenberg (2013) define it as: ‘the capacity of individuals to navigate their ways to resources that sustain
4 well-being. The capacity of individuals’ physical and social ecologies to provide those resources, and the capacity of individuals, their families and their communities to negotiate culturally meaningful ways to share resources’ (p. 3).
It is important to keep in mind that resilience should not be conceived as a direct indicator of individual differences in well-being. For example, resilient (but high risk) children were found to be comparable in developmental outcome with lower-risk children (Fraser et al., 1999). Resilience is a mechanism affecting both protective and risk factors. Hence, it is more than the accumulation of protective factors (Ungar, 2011). It positively affects the balance between risk and protective factors, not by reducing stress or risks, but by ensuring that an individual can deal with adversity effectively (Werner, 2000). Fraser et al. (1999) argue that unexposed children to high risk situations or adversity may not develop resilience. However, this can be disproved since resilience is not a characteristic of the individual himself, but of the individual and its environment. It is made measurable by focusing on the social context of an individual (Ryan, 2001).
Gender differences in resilience
Various studies indicate that females are more resilient than males (Hsieh & Shek, 2008; Roberts, 2017). The study of Roberts (2017) on abused females cautiously concluded that they are more likely to show resilience. Besides, this study suggests that girls, rather than boys, are influenced more positively by interpersonal protective factors, such as parent and peer support A study by Van der Put et al. (2014) showed that risks in the interpersonal domain affect girls stronger. It therefore seems that the interpersonal domain more strongly affects girls than boys, both in terms of risk and protective factors, which might result in gender differences in resilience. Hsieh and Shek (2008) concluded that females possess higher academic and personal resilience than males. Contrary to previous studies, Santos (2017) found that males showed
5 significantly more resilience than females, but they did not find a clear explanation why males scored higher.
To conclude, research on differences in resilience between males and females yields equivocal results (Hjemdal, Vogel, Solem, Hagen, & Stiles, 2011; Roberts, 2017; Santos, 2017; Shek, Chin & Lin, 2016). However, the questionnaires used to assess resilience in these studies have not been tested for measurement invariance. Therefore, it is not clear whether results are driven by gender bias or reflect true differences in resilience between females and males. This requires the determination of measurement invariance of resilience across gender.
Education and resilience
The literature on education level and resilience is rare, especially in youth. Some evidence is found for an association between education level and resilience (Campbell-Sills, Forde, & Stein, 2009; Frankenberg, Sikoki, Sumantri, Suriastini, & Thomas, 2013). For example, higher educated people who experienced a natural disaster are more likely to be resilient than people with lower education (Frankenberg et al., 2013). In a sample of people who did not experience a natural disaster, people seemed more resilient for stress related experiences if they attended higher education instead of lower levels of education (Campbell-Sills et al., 2009). In contrast, studies have found that among adult females and males education level was not associated with resilience (Scali, et al. 2012; Wells, 2009). Some studies examined the relation between support received at school and resilience of students. However, no differences between education level and the extent of resilience in youth have been researched so far (Brodersen, 2013; Liebenberg, Ungar, & LeBlanc, 2013). The equivocal results emphasize the importance of investigating measurement invariance to establish whether differences in resilience scores on questionnaires are affected by differences in response tendencies of research participants having attended
6 different levels of education. This may affect the factor structure of the instrument that is used to examine resilience.
Psychopathology and resilience
Literature describes the relation between resilience and psychopathology (Arslan, 2016; Edward, 2005; Hjemdal, Aune, Reinfjell, Stiles, & Friborg, 2007; Hjemdal et al., 2011; Skrove, Romundstad, & Indredavik, 2013). For example, the study of Arslan (2016) concluded that psychological maltreatment, which is associated with higher levels of psychopathology, negatively correlates with resilience. In other studies, researchers found that resilience reduces the chance of becoming depressed or stressed (Edward, 2005; Hjemdal et al., 2007; Skrove et al., 2013). Hjemdal et al. (2011) measured resilience with personal dispositions, the availability of sources of social support outside the family and family cohesion. They showed that increased resilience scores correlated with lower levels of stress and psychiatric symptoms. To examine concurrent validity, psychopathology symptoms were correlated with the extent of resilience as measured with the CYRM-12.
Previous validation of the CYRM-12
The CYRM-12 is a shortened version of the CYRM-28. The CYRM-28 was developed in the International Resilience Project. The CYRM-28 was originally designed to measure resilience with youth aged 9 to 23 years (Ungar & Liebenberg, 2013). The first validation study was conducted by Liebenberg (Liebenberg, Ungar, & Vijver, 2012). They performed an exploratory factor analysis (EFA) in a mixed gender sample and found a three-factor solution. Firstly, the individual factor consists of personal skills, peer support and social skills. Secondly, the caregiving factor consists of physical caregiving and psychological caregiving. Thirdly, the context factor consists of spiritual, educational and cultural resources. Subsequently, a
7 confirmative factor analysis (CFA) corroborated the same three factors. In addition, the researchers found that females and youth with ethnic minority status scored higher on resilience than males and youth with majority status. However, the effect size for gender differences was small compared to the larger effect size for minority/majority status. This first validation study of the CYRM-28 had acceptable Cronbach’s alpha reliabilities and high intraclass correlations (agreement between assessors).
In 2013 Liebenberg et al. (2013) conducted a validation study on a shortened version of the CYRM, namely, the CYRM-12. In this study, the researchers used two mixed gender samples to test the validity of the questionnaire. The first (construction) sample was used for an EFA to select items for a shortened version. The researchers found a four-factor solution with twelve items. The second (validation) sample was used for a CFA. In both samples the Cronbach’s alpha reliability was satisfactory, and in the second sample the fit was also satisfactory. After this first validation study of the CYRM-12, more articles were published (Arslan, 2015; Mu & Hu, 2016; Panter-Brick, et al. 2018).
A study carried out in Jordan, consisting of both males and females, investigated convergent validity (correlation between results of current and new research) of the CYRM-12 (Panter-Brick et al. 2018). They conducted a CFA to shorten the CYRM-28 and found a three-factor solution with different items than the original CYRM-12. The study showed a negative association between resilience and mental health symptoms. A Chinese mixed gender validation study of the CYRM-12 revealed a single factor solution in CFA. The CYRM-12 showed high internal consistency reliability in terms of Cronbach’s alpha (Mu & Hu, 2016). Arslan (2015) examined the reliability and validity of the CYRM-12 through EFA and CFA, and found a single factor solution. Convergent validity was established through a positive correlation between self-efficacy and resilience. The Cronbach’s alpha reliability was high. None of the validation studies of the CYRM-12 investigated measurement invariance.
8 Current study
The aim of this study was to examine the internal structure, reliability and concurrent validity of the CYRM-12 for Dutch adolescents between the ages of 14 and 22. Measurement invariance was tested for gender and education level. Concurrent validity of the CYRM-12 was examined by testing the correlation between psychopathology, measured with The Symptom CheckList (SCL-90-R), and resilience
Method Participants
The sample consisted of N = 763 Dutch students (21.9% male and 78.1% female) between 16 and 24 years old (M = 20.10 year; SD = 2.55 year). Most participants were Caucasian white (88.2%), while 10.4% did have a Surinamese background. Suriname is one of the former colonies of The Netherlands with an ethnic diverse population. The remaining participants did have a Dutch Antillean, Turkish or other ethnic background. Students attended vocational (45.0%) or academic education (55.0%).
Procedure
Convenience sampling was used to recruit participants, partly through snowball sampling. A nation-wide campaign marked the start of the recruitment through social media, three higher education colleges, several performances on radio, YouTube and television. A flyer was used on the different schools. Students applied for participation and filled out the questionnaires online with a smartphone (www.g-moji.nl). The goal of the study was explained by text and video-tutorials. Additional questions could be asked directly to the research team. To ensure that every participant was aware of their rights and our privacy statement, they all signed an informed consent. Participation was voluntary and termination was always possible. As a reward, the participants received a podcast series of four episodes with the story of an
9 adolescent experiencing mental health problems and receiving psychiatric care. Inclusion and exclusion criteria were based on the type of smartphone operating system and whether the smartphone use was work-related or for personal use. The G-Moji app is only available for Android, so participants with other operating systems (e.g., iOS) were excluded. In total 964 adolescents applied with a non-response rate of 20.85%.
Instruments
The CYRM-12 was used to measure resilience, which is a shortened version of the CYRM-28 (Liebenberg et al., 2013). The questionnaire consists of 12 items rated on a 5-point Likert type scale (1 = not at all, 5 = a lot); see appendix A. The CYRM-12 is also available with a 3-point Likert type scale, but the 5-point scale was used in the present study.
The SCL-90-R is a questionnaire with 90 items rated on a 5-point Likert scale (0 = not at all, 4 = extremely) measuring a range of psychological problems and symptoms of psychopathology, including somatization, obsessive compulsive behavior, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation and psychoticism (Martinez, Stillerman, & Waldo, 2005). The Global Severity Index reflects the total score of psychopathological symptoms. Examples of items are: ‘Nervousness or shakiness inside’, ‘Trouble remembering things’ (Eich, Angst, Frei, Ajdacic-Gross, Rössler, & Gamma, 2012). The SCL-90-R demonstrated favorable psychometric properties, such as convergent and concurrent validity, as well as high reliability (Martinez et al., 2005; Schmitz et al., 2000; Vallejo, Jordán, Díaz, Comeche, & Ortega, 2007). Internal consistency reliability of the overall psychopathology score in the present study in terms of Cronbach’s alpha was high (α = .98; Field, 2018).
10 Statistical Analysis
The data were entered in IBM SPSS Statistics 24 (SPSS). There were no missing values. Firstly, a Principle Component Analysis (PCA) with oblimin rotation was conducted to examine whether a one-, two- or three-factor structure best fitted the data. The number of factors was based on the structure’s eigenvalue if higher than one (Kaiser, 1960). After the PCA, a CFA was conducted in R x64 3.6.0 (R). Firstly, to examine the fit of the PCA model and subsequently to test measurement invariance of the CYRM-12 across gender and education (Gregorich, 2006; Meredith, & Teresi, 2006). Modification indices were consulted to improve model fit. Residual errors of items with similar wordings or comparable content were allowed to covary. Next, items had to have a factor loading above .30 (Peterson, 2000). The following types of invariances were examined: configural invariance indicates that common factors are associated with the same items across gender and education. Metric invariance indicates that factor loadings are equal across groups, meaning that common factors have the same meaning across gender and education level. Scalar invariance indicates that intercepts are equal across groups, and comparisons of means between male and female adolescents or different education levels are meaningful. At last, strict invariance was tested to check if the factor loadings, intercepts and residuals variances were the same across gender and education.
The same fit-indexes were used for the CFA and the measurement invariance of the CYRM-12: the Comparative Fit Index (CFI), the Tucker-Lewis Index (TLI), the root mean square error of approximation (RMSEA) and the standardized root mean square residual (SRMR; Kline, 2016). A CFI value larger than 0.95 and TLI value close to .95 represent a good factor model fit (Hu & Bentler, 1999). A RMSEA lower than .05 and SRMR lower than .06 represent a good factor model fit. If the items were not normally distributed, the MLR-estimator would be used, which is robust for violations of normality, see appendix B. Scalar invariance
11 should be established to conduct a independent sample t-test to compare mean resilience scores between male and female adolescents or vocational and academic education.
Finally, the reliability was calculated for the CYRM-12 with Cronbach’s Alpha. The reliability is considered sufficient above .60 and good above .80 (Field, 2018). After calculating the Cronbach’s Alpha, Pearson r between the sum scores of the SCL-90-R and the CYRM-12 was calculated. Pearson r value below .10 indicates a weak association, above .30 a moderate association and above .50 a strong association (Van Peet, Van den Wittenboer, & Hox, 2004).
Results
First, a PCA was conducted to examine the factor structure of the CYRM-12. The PCA yielded a two-factor structure, explaining 28.32% of the variance. However, the two factors could not be meaningfully interpreted in accordance with the original factor structure (individual, caregiving and context) of the CYRM-28. A one-factor solution showed that item 1’I have people I look up to’ had a factor loading below .30 and therefore was removed from by the PCA.
Next a CFA was conducted in R. In Table 2 the fit statistics of the models are presented. The MLR-estimator was used, because items were not normally distributed, see appendix B. Model 1 was tested with the 11 items that were derived from the PCA, but this model fitted the data poorly. Subsequently, model 2 was tested, with the same 11 items, of which items 3 and 8, 2 and 4, and 2 and 7 were allowed to correlate. This model also showed insufficient fit. In Model 3, item 2 was deleted because the factor loading was below .30 (.256), and it showed high correlations with the items 4 and 7. The correlated errors were removed from this model, but still the model fit proved to be unsatisfactory. In the last Model 4, item 2 was removed, and the residual error variance of items 3 and 8 were allowed to covary (r = .404). The fourth and final model showed a good fit to the data, with the following fit indexes: χ2(34) = 89.834, p < .001; RMSEA = .048; TLI = .938; CFI = .953; SRMR = .038. Ten items had a factor loading
12 of at least .320 or above, see Table 1. The Cronbach’s Alpha was acceptable for the CYRM-12 with ten items (α = .770).
Measurement invariance was examined by comparing male and female adolescents and different education levels. The fit statistics of the measurement invariance models of gender and education level are presented in Table 3 and Table 4. Strict invariance was established for gender according to the CFI criterion, although the chi-square test showed a significant drop in model fit for scalar invariance (p < .05). However, this test depends on sample size, hence, was not used as a criterion of model fit. A t-test showed no significant difference in mean resilient
Table 1
CYRM-12: The Best Fitting CFA Model
Item Description Factor loading
3. My parent(s)/caregiver(s) know a lot about me .413
4. I try to finish what I start .320
5. I am able to solve problems without harming myself or others (for example by using drugs and/or being violent)
.453 6. I know where to go in my community to get help .565
7. I feel I belong at my school .562
8. My family stands by me during difficult times .532
9. My friends stand by me during difficult times .461
10. I am treated fairly in my community .631
11. I have opportunities to develop skills that will be useful later in life (like job skills and skills to care for others)
.556
12. I enjoy my community’s traditions .447
Note. From Ungar, M. and Liebenberg, L., 2013, the user’s manual “The Child and Youth Resilience Measure
(CYRM) Youth Version”, p. 17. Resilience Research Centre.
Table 2
CFA: Fit indexes of CYRM-12
X2 DF p RMSEA TLI CFI SRMR
CYRM-12
Model 1 236.172 44 <.001 .080 .810 . 848 .057 Model 2 105.083 41 <.001 .048 .932 . 949 .039 Model 3 172.792 35 <.001 .075 .849 .883 .050 Model 4 89.834 34 <.001 .048 .938 .953 .038
13 scores between male and female adolescents (t(761) = .675, p = .500). Next, only metric invariance was established for education level according to the CFI criterion, with also a significant drop in model fit (p < .001), so means of different education levels could not be compared.
Table 3
CFA: Measurement invariance for gender
X2 DF p RMSEA TLI CFI SRMR
Gender Configural 112.419 68 <.001 .043 .950 .962 .039 Metric 123.462 77 <.001 .041 .954 .961 .044 Scalar1 142.622 86 <.001 .043 .951 .953 .047 Strict 153.855 96 <.001 .040 .956 .953 .047 Table 4
CFA: Measurement invariance for education
X2 DF p RMSEA TLI CFI SRMR
Education
Configural 105.326 68 <.001 .039 .954 .965 .038 Metric 115.342 77 <.001 .037 .959 .965 .043 Scalar2 151.099 86 <.001 .046 .938 .940 .053 Strict 201.121 96 <.001 .054 .913 .907 .065
At last concurrent validity was examined with the sum scores of the SCL-90-R and the CYRM-12, showing a high negative association between psychopathological symptoms and resilience (r = -.579, p < .001), which provides evidence for concurrent validity.
1χ2 (9) = 19.160, p = .024
14 Discussion
The aim of this study was to investigate the internal structure, reliability and concurrent validity of the CYRM-12 among Dutch adolescents. In addition, the measurement invariance was investigated for gender and education. A PCA yielded a one-factor solution, which was confirmed in a CFA. Two items were removed in the CFA as they showed high correlations with other items or had a low factor loading. Reliability was acceptable and evidence for concurrent validity was found in a strong negative association between resilience and psychopathology. Strict measurement invariance was established for gender and metric invariance for education level. No significant differences in resilience were found between male and female adolescents.
This was the first study in the Netherlands to investigate the internal structure, reliability and validity of the CYRM-12. The results implicate that resilience can be validly and reliably measured among Dutch adolescents with the shortened version of the CYRM. The one-factor structure is not in line with results of the validation studies of the CYRM-12 by Liebenberg et al. 2013 and Panter-Brick et al. (2018). Looking at the Jordan study, the disparities may be due the differences in used items for the shortened version (Panter-Brick et al., 2018). Studies conducted in China and Turkey found a one-factor structure (Arslan, 2015; Mu & Hu, 2016), which is in line with results from the current study. Two items of the CYRM-12 were deleted. Notably, the remaining ten items tapped the three dimensions (individual, caregiving and context) of the original CYRM-283. Hence, this means the shorted Dutch 10-item version of the CYRM does not underrepresent the concept of resilience, and may be considered a useful instrument.
Previous studies on resilience did not examine measurement invariance for gender, but still examined gender differences in resilience. The study from Santos (2017) revealed that
15 males showed more resilience then females. However, many studies found that females were more resilient than males (Hjemdal et al., 2011; Roberts, 2017; Shek et al., 2016). Our study, which established strict measurement invariance across male and female adolescents, did not find a difference in resilience. Rather, it suggests that gender differences in resilience may be caused by lack of measurement invariance. Therefore, future research outside The Netherlands should first establish measurement invariance of instrument before testing gender differences while assessing resilience.
Only metric invariance was found for education level. Notably, items yielded small to medium differences between academic education compared to vocational education. Item 3 ‘My parent(s)/caregiver(s) know a lot about me’ and item 12 ‘I enjoy my community’s traditions’ showed small differences between the educational groups. Meanwhile, item 11 ‘I have opportunities to develop skills that will be useful later in life (like job skills and skills to care for others)’ showed a large difference between academic and vocational education. These differences explain why no scalar or strict invariance were found. The large difference for item 11 can be explained by the examples that are shown in brackets, in particular with respect to ‘job skills’. This item refers to useful skills that can be acquired at school, and may be affected by attending academic or vocational education.
A strong relation between psychopathological symptoms and a lack of resilience was found, supporting concurrent validity of the CYRM. This is in line with results from previous studies (Arslan, 2016; Edward, 2005; Hjemdal et al., 2007; Hjemdal et al., 2011; Skrove et al., 2013). This is an essential outcome for clinical practice, because treatment that enhances resilience, which can be validly and reliably monitored with the Dutch brief version of the CYRM, may result in lower levels of psychopathology.
The current study showed a few limitations. First, measurement invariance for culture could not be tested, because few participants with a migration background were present.
16 Secondly, convenience sampling was used. It is therefore unknown whether the sample adequately represents the research population of interest (Maruyama & Ryan, 2014). Thirdly, test-retest reliability was not investigated (Maruyama, & Ryan, 2014). Lastly, convergent, predictive and discriminant validity were not investigated.
Despite these limitations, the current study demonstrated that the Dutch shortened version of the CYRM has favorable psychometric properties. Further research would be necessary to investigate validity and reliability of the Dutch shortened ten item version of the CYRM in clinical groups and youth with a migration background. As mentioned before, different cultures should be compared. Tests are not always appropriate for every cultural background (Miller-Jones, 1989), thus for multicultural countries as the Netherlands this may impose issues. In addition, a clinical sample and a non-clinical sample should be compared.
17 References
Arslan, G. (2015). Psychometric properties of Child and Youth Resilience Measure (CYRM-12): The study of reliability and validity. Ege Eğitim Dergisi, 16(1), 1-12.
Arslan, G. (2016). Psychological maltreatment, emotional and behavioral problems in adolescents: The mediating role of resilience and self-esteem. Child Abuse & Neglect, 52, 200-209. https://doi.org/10.1016/j.chiabu.2015.09.010
Brodersen, E. M. (2013). The Child and Youth Resilience Measure in an Adolescent Offender
Population (Doctoral dissertation, Arts & Social Sciences: Department of Psychology).
Campbell-Sills, L., Forde, D. R., & Stein, M. B. (2009). Demographic and childhood environmental predictors of resilience in a community sample. Journal of psychiatric
research, 43(12), 1007-1012. https://doi.org/10.1016/j.jpsychires.2009.01.013
Dam, L. V., Rietstra, S., Drift, E. V. D., Stams, G. J., Mei, R. V. D., Mahfoud, M., ... & Reid, T. (2019). Can an emoji a day keep the doctor away? An explorative mixed-methods feasibility study to develop a self-help app for youth with mental health problems. Frontiers in psychiatry, 10, 593. https://doi.org/10.3389/fpsyt.2019.00593 Dumont, M., & Provost, M. A. (1999). Resilience in adolescents: Protective role of social
support, coping strategies, self-esteem, and social activities on experience of stress and depression. Journal of youth and adolescence, 28(3), 343-363. https://doi.org/10.1023/A:1021637011732
Edward, K. L. (2005). Resilience: A protector from depression. Journal of the American
Psychiatric Nurses Association, 11(4), 241-243.
https://doi.org/10.1177/1078390305281177
Eich, D., Angst, J., Frei, A., Ajdacic-Gross, V., Rössler, W., & Gamma, A. (2012). A new rating scale for adult ADHD based on the Symptom Checklist 90 (SCL-90-R). European
18
archives of psychiatry and clinical neuroscience, 262(6), 519-528.
https://doi.org/10.1007/s00406-011-0288-1
Field, A. (2018). Discovering Statistics Using IBM SPSS Statistics (5). Chapter 18: Exploraty factor Analysis. (pp. 989-1061) London, United Kingdom: SAGE.
Frankenberg, E., Sikoki, B., Sumantri, C., Suriastini, W., & Thomas, D. (2013). Education, vulnerability, and resilience after a natural disaster. Ecology and society: a journal of
integrative science for resilience and sustainability, 18(2), 16. doi:
10.5751/ES-05377-180216
Fraser, M. W., Galinsky, M. J., & Richman, J. M. (1999). Risk, protection, and resilience: Toward a conceptual framework for social work practice. Social work research, 23(3), 131-143. https://doi.org/10.1093/swr/23.3.131
Gregorich, S. E. (2006). Do self-report instruments allow meaningful comparisons across diverse population groups? Testing measurement invariance using the confirmatory factor analysis framework. Medical care, 44(11 Suppl 3), S78. doi: 10.1097/01.mlr.0000245454.12228.8f
Hjemdal, O., Aune, T., Reinfjell, T., Stiles, T. C., & Friborg, O. (2007). Resilience as a predictor of depressive symptoms: a correlational study with young adolescents. Clinical Child
Psychology and Psychiatry, 12(1), 91-104. https://doi.org/10.1177/1359104507071062
Hjemdal, O., Vogel, P. A., Solem, S., Hagen, K., & Stiles, T. C. (2011). The relationship between resilience and levels of anxiety, depression, and obsessive–compulsive symptoms in adolescents. Clinical psychology & psychotherapy, 18(4), 314-321. https://doi.org/10.1002/cpp.719
Hsieh, M. O., & Shek, D. T. (2008). Personal and family correlates of resilience among adolescents living in single-parent households in Taiwan. Journal of Divorce & Remarriage, 49(3-4), 330-348. https://doi.org/10.1080/10502550802221925
19 Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural equation modeling: a
multidisciplinary journal, 6(1), 1-55. https://doi.org/10.1080/10705519909540118
Kaiser, H. F. (1960). The application of electronic computers to factor analysis. Educational and Psychological Measurement, 20, 141–151.
https://doi.org/10.1177/001316446002000116
Kline, R.B. (2016). Principles and practice of structural equation modeling (4th ed.). New York, NY: Guilford Press.
Liebenberg, L., Ungar, M., & LeBlanc, J. C. (2013). The CYRM-12: A brief measure of resilience. Canadian Journal of Public Health, 104(2), e131-e135. https://doi.org/10.1007/BF03405676
Liebenberg, L., Ungar, M., & Vijver, F. V. D. (2012). Validation of the child and youth resilience measure-28 (CYRM-28) among Canadian youth. Research on social work
practice, 22(2), 219-226. https://doi.org/10.1177/1049731511428619
Martinez, S., Stillerman, L., & Waldo, M. (2005). Reliability and validity of the SCL-90-R with Hispanic college students. Hispanic Journal of Behavioral Sciences, 27(2), 254-264. https://doi.org/10.1177/0739986305274911
Maruyama, G., & Ryan, C. S. (2014). Research methods in social relations. Chapter 7: Measurement and Reliability. (pp. 163-200) John Wiley & Sons.
Maruyama, G., & Ryan, C. S. (2014). Research methods in social relations. Chapter 9: Sampling Methods. (pp. 231-253) John Wiley & Sons.
Meredith, W., & Teresi, J. A. (2006). An essay on measurement and factorial invariance. Medical care, S69-S77.
20 Mu, G. M., & Hu, Y. (2016). Validation of the Chinese version of the 12-item child and youth resilience measure. Children and youth services review, 70, 332-339. https://doi.org/10.1016/j.childyouth.2016.09.037
Panter‐Brick, C., Hadfield, K., Dajani, R., Eggerman, M., Ager, A., & Ungar, M. (2018). Resilience in context: A brief and culturally grounded measure for Syrian refugee and Jordanian host‐community adolescents. Child development, 89(5), 1803-1820. https://doi.org/10.1111/cdev.12868
Peterson, R. A. (2000). A meta-analysis of variance accounted for and factor loadings in exploratory factor analysis. Marketing Letters, 11(3), 261-275. https://doi.org/10.1023/A:1008191211004
Roberts, D. M. (2017). Resilience in physically maltreated adolescents: Interpersonally related
protective factors and gender differences (Doctoral dissertation, The George Washington
University).
Ryan, D. P. J. (2001). Bronfenbrenner’s ecological systems theory. Retrieved February, 25, 2020.
Santos, J. K. (2017). Peer Victimization, Self-esteem and Social Anxiety as Predictors of Resilience: Gender Differences in Resilience Explored (Master’s thesis). University of Chester, United Kingdom.
Scali, J., Gandubert, C., Ritchie, K., Soulier, M., Ancelin, M. L., & Chaudieu, I. (2012). Measuring resilience in adult women using the 10-items Connor-Davidson Resilience Scale (CD-RISC). Role of trauma exposure and anxiety disorders. PloS one, 7(6). Schmitz, N., Hartkamp, N., Kiuse, J., Franke, G. H., Reister, G., & Tress, W. (2000). The
symptom check-list-90-R (SCL-90-R): a German validation study. Quality of Life
21 Shek, D. T. L., Chi, X., & Lin, L. (2016). Change and correlates of resilience in high school students in hong kong: Findings based on six waves of longitudinal data. International Journal on Disability and Human Development, 15(4), 387-396. doi:10.1515/ijdhd-2017-5006
Skrove, M., Romundstad, P., & Indredavik, M. S. (2013). Resilience, lifestyle and symptoms of anxiety and depression in adolescence: the Young-HUNT study. Social psychiatry and
psychiatric epidemiology, 48(3), 407-416. https://doi.org/10.1007/s00127-012-0561-2
Strolin-Goltzman, J., Woodhouse, V., Suter, J., & Werrbach, M. (2016). A mixed method study on educational well-being and resilience among youth in foster care. Children and youth
services review, 70, 30-36. https://doi.org/10.1016/j.childyouth.2016.08.014
Ungar, M. (2011). The social ecology of resilience: Addressing contextual and cultural ambiguity of a nascent construct. American Journal of Orthopsychiatry, 81(1), 1. https://doi.org/10.1111/j.1939-0025.2010.01067.x
Ungar, M., & Liebenberg, L. (2013). Child & Youth Resilience Measure (CYRM) Youth Measure: User’s Manual. Resilience Research Centre.
Vallejo, M. A., Jordán, C. M., Díaz, M. I., Comeche, M. I., & Ortega, J. (2007). Psychological assessment via the internet: a reliability and validity study of online (vs paper-and-pencil) versions of the General Health Questionnaire-28 (GHQ-28) and the Symptoms Check-List-90-Revised (SCL-90-R). Journal of Medical Internet Research, 9(1), e2.
doi: 10.2196/jmir.9.1.e2
Van der Put, C.E., Deković, M., Hoeve, M., Stams, G.J.J.M., Van der Laan, P.H., & Langewouters, F.E.M. (2014). Risk assessment of girls: are there any sex differences in risk factors for reoffending and in risk profiles? Crime and Delinquency, 60 (7), 1033-1056. doi: 10.1177/0011128710384776
22 Van Peet, A. A. J., Van den Wittenboer, G. L. H., & Hox, J. J. (2004), Effectgrootte en het
onderscheidingsvermogen van een toets. In A. A. J van Peet, G. L. H. van den
Wittenboer, & J. J. Hox, Toegepaste statistiek: beschrijvende technieken. Groningen: Wolters Noordhof.
Wells, M. (2009). Resilience in rural community‐dwelling older adults. The Journal of Rural
Health, 25(4), 415-419. https://doi.org/10.1111/j.1748-0361.2009.00253.x
Werner, E. E. (2000). Protective factors and individual resilience. Handbook of early childhood
23 Appendix A
The Child and Youth Resilience Measure-12 (Dutch translation) 1. Ik heb mensen waar ik tegenop kijk
2. Een opleiding afronden is belangrijk voor mij 3. Mijn ouder(s)/verzorger(s) weten veel over mij 4. Ik probeer af te maken waar ik aan begin
5. Ik ben in staat problemen op te lossen zonder mezelf of anderen schade toe te brengen (bijv. door drugsgebruik en/of geweld)
6. Ik weet waar ik in mijn omgeving hulp kan krijgen 7. Ik voel(de) me thuis op school
8. Mijn familie steunt me in moeilijke tijden 9. Mijn vrienden steunen me in moeilijke tijden 10. Ik word eerlijk behandeld in mijn omgeving
11. Ik heb de kans om vaardigheden te ontwikkelen die later van pas komen (zoals voor werk en om te zorgen voor anderen)
Appendix B
Descriptive Statistics of all CYRM-12 items
v1 v2 v3 v4 v5 v6 v7 v8 v9 v10 v11 v12 Valid 763 763 763 763 763 763 763 763 763 763 763 763 Missing 0 0 0 0 0 0 0 0 0 0 0 0 Mean 2.866 4.229 3.364 3.776 3.794 3.418 2.929 3.433 3.450 3.482 3.830 3.165 Median 3.000 5.000 4.000 4.000 4.000 4.000 3.000 4.000 4.000 4.000 4.000 3.000 Mode 2.000 5.000 4.000 4.000 5.000 4.000 4.000 4.000 4.000 4.000 4.000 4.000 Std. Deviation 1.076 1.079 1.155 1.071 1.201 1.224 1.291 1.249 1.224 1.122 1.068 1.194 Skewness 0.236 -1.535 -0.338 -0.694 -0.680 -0.300 -0.037 -0.306 -0.370 -0.431 -0.776 -0.076 Std. Error of Skewness 0.089 0.089 0.089 0.089 0.089 0.089 0.089 0.089 0.089 0.089 0.089 0.089 Kurtosis -1.074 1.531 -1.018 -0.644 -0.839 -1.188 -1.209 -1.175 -1.098 -1.035 -0.432 -1.266 Std. Error of Kurtosis 0.177 0.177 0.177 0.177 0.177 0.177 0.177 0.177 0.177 0.177 0.177 0.177 Shapiro-Wilk 0.862 0.707 0.869 0.807 0.816 0.861 0.895 0.869 0.864 0.832 0.811 0.863 P-value of Shapiro-Wilk < .001 < .001 < .001 < .001 < .001 < .001 < .001 < .001 < .001 < .001 < .001 < .001 Minimum 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Maximum 5.000 5.000 5.000 5.000 5.000 5.000 5.000 5.000 5.000 5.000 5.000 5.000
