Factor structure and psychometric properties of the Hopkins Symptom Checklist: An investigation with culturally and linguistically diverse youth in Australia

Abstract

Background

The factor structure and psychometric properties of the Hopkins Symptom Checklist (HSCL‐30), a derivative of the symptom checklist developed by Derogatis et al. in 1974, is investigated for the first time in a sample of Australian culturally and linguistically diverse (CALD) youth.

Methods

CALD youth (N = 241) completed the checklist with the help of an interpreter. An exploratory factor analysis was conducted and psychometric properties tested.

Results

Compared to the longer version, a 16‐item unidimensional factor structured emerged as the best solution. It was labelled as Hopkins Symptom Checklist–CALD Youth (HSCL‐Y). The items are internally consistent (α = .91). Convergent, divergent and discriminant validity is supported. The HSCL‐Y positively correlated with measures of emotional symptoms and acculturative stress. The 16‐item measure was negatively correlated with measures of well‐being and differentiated a sample with high emotional reaction from those who did not endorse any emotional experience.

Conclusion

The new HSCL‐Y was found to be a psychometrically sound instrument in this sample of CALD youth, with potential clinical utility in school samples to screen for psychological distress.

Key words:

• Australia
• culture
• diversity
• Hopkins Symptom Checklist
• psychometric properties

WHAT IS ALREADY KNOWN ON THIS TOPIC?

• Culturally and linguistically diverse youth can experience psychological distress.

• It is important to screen their psychological distress in order to assist them in the best possible manner.

• Presently, there is a dearth of valid screening measures that can be used with culturally and linguistically diverse youth.

WHAT THIS PAPER ADDS?

• A 16‐item Hopkins Symptoms Checklist is an unidimensional measure.

• This shorter checklist is reliable and a valid measure.

• It can be used to assess the psychological distress of culturally and linguistically diverse youth.

International migration has reached an unprecedented peak with the number of migrants and refugees reaching 258 million people in 2017 (United Nations, 2017). Two‐thirds of these individuals live in 20 developed countries in the West, including Australia. Thirty‐one million relocated people are minors, moving to a developed country with a culture and a language very different from their country of origin (UNICEF, 2017). They leave their countries of origin either to flee from aversive and life‐threatening circumstances or for better life prospects (Betancourt, Yudron, Wheaton, & Smith‐Fawzi, 2012). Subsequently, minors from migrant or refugee backgrounds become part of the culturally and linguistically diverse (CALD) populations in Western nations (Australian Bureau of Statistics, 2012). Migration can be stressful (Birman, Simon, Chan, & Tran, 2014) and a combination or pre‐, peri‐, or post‐migration stressors can precipitate psychological distress (Carswell, Blackburn, & Barker, 2011; Hocking, Kennedy, & Sundram, 2015; Nickerson et al., 2015). The ramifications for relocated CALD youth include an increased risk of depression, anxiety, and somatisation symptoms (Amone‐P'Olak, Garnefski, & Kraaij, 2007; Grgic et al., 2005). Due to the heightened vulnerability of this population and the increasing numbers of CALD youth in Australian society, a valid and reliable screening tool for psychological distress is of paramount importance for identifying at‐risk individuals.

CALD youth may confer an increased risk of psychological distress due to the interplay of exposed stressors (Tempany, 2009). Studies have indicated the impact of pre‐, peri‐, and post‐migration stressors such as war, trauma, multiple losses (family, friends, culture, ethnic identity), acculturation, social exclusion, education, discrimination, and intergenerational conflict on youth (Fazel, Reed, Panter‐Brick, & Stein, 2012; Pacione, Measham, & Rousseau, 2013). Previous studies, identifying at‐risk individuals and investigating ways to mediate these risks, result in mixed findings (Fazel et al., 2012). Potential risk factors for anxiety, depression, and somatisation symptomology include female gender, older age (e.g., 15-years and older), ethnic origin, and refugee status (Fazel et al., 2012). Half of the studies in the literature note a higher prevalence of internalising pathology in girls compared to that in boys, thus indicating mixed findings (Derluyn, Mels, & Broekaert, 2009; Hodes, Jadev, Chandra, & Cuniff, 2008; Renner & Salem, 2009). Older age of CALD youth has also been considered a potential risk factor, with previous studies noting that older adolescents (aged 13–17-years) reported elevated depression and stress scores compared to younger individuals (Hasanovic, Sinanovic, Selimbasic, Pajevic, & Avdibegovic, 2006; Yurtbay, Alyanak, Abali, Kaynak, & Durukan, 2003). Other studies found no independent relationship between psychological symptoms and age of youth (Ellis et al., 2016). Studies investigating the relationship between psychological distress and ethnic origin report mixed findings. Hodes et al. (2008) found that ethnic origin did not affect scores on a measure of trauma. However, refugee youth originating from the Middle East had elevated depression scores compared to other youth. Interestingly, Bean, Derluyn, Eurelings‐Bontekoe, Broekaert, and Spinhoven (2007) found higher externalising and internalising scores (on Hopkins Symptom Checklist [HSCL]‐37A) on children from Guinea, Ethiopia, and Eritrea compared to children from China and other African countries. In contrast, Tousignant et al. (1999) found that children, from the Middle East, scored lower on psychological distress compared to all other regions assessed. Despite unclear outcomes, there is evidence that varying degrees of trauma and stress may aggravate risk and can have a long‐term impact on the psychosocial development of children (Vervliet, Lammertyn, Broekaert, & Derluyn, 2014). Future research on the nature of the relationship between psychological functioning and psychosocial well‐being is warranted.

Early intervention requires early identification of emotional risk factors, yet we have limited access to psychometrically valid methodologies for the assessment of psychosocial risk factors such as depression, anxiety, and somatisation. The development of psychometrically valid methodologies has potential to contribute to school‐based culturally appropriate interventions to promote CALD youth's post‐migration adjustments (Beehler, Birman, & Campbell, 2012; Mitchelson et al., 2010; Quinlan, Schweitzer, Khawaja, & Griffin, 2016). Despite these developments, there is a scarcity of valid screening measures that can be used with CALD youth.

Another consideration in identifying a clinically useful screening tool to measure psychological distress in this cohort is the representation of the construct itself. In Western cultures, the idea of mind–body duality is a central feature in health and mental health (Scheper‐Hughes & Lock, 1987; Wen, 1998). However, research and clinical experiences of professionals suggest this differentiation of mental health is much more diffuse in non‐Western cultures. The paucity of research into the number of latent factors in symptom checklists, and attitudes and beliefs of mental health, provides support for the view of mind–body holism in many non‐Western cultures (Scheper‐Hughes & Lock, 1987). Further, research has illustrated a strong link between stigma and socially desirable responses on self‐report questionnaires measuring symptoms such as anxiety and depression (Whitley, Kirmayer, & Groleau, 2006; Wu, Kviz, & Miller, 2009). Similarly, it has been intimated that culture interacts with the manifestation of psychological distress (Syed, Zachrisson, Dalgard, Dalen, & Ahlberg, 2008; Vonnahme, Lankau, Ao, Shetty, & Cardozo, 2015). For instance, in cultures where is it unacceptable to verbally express distress, research has found individuals (notably youth) are more likely to somatise than to express mood‐related symptoms associated with depression (Vonnahme et al., 2015; Bean et al., 2007). The impact of linguistics also plays a role in reliably and validly assessing mental well‐being of CALD youth. Therefore, it is essential to find a measure that takes these cultural considerations into account, and can reliably and validly assess for psychological distress in this sample. The Hopkins Symptom Checklist (HSCL) is a widely used instrument in clinical settings and more recently in refugee, asylum seeker, and migrant samples of adults and children. Although most of the research on this scale was conducted in Western and adult cohorts its utility with CALD youth has not been investigated.

The HSCL is a symptom checklist originally developed as an outcome measure (Parloff, Kelman, & Frank, 1954). The HSCL has undergone multiple revisions, resulting in various versions varying from 25 to 90 items. The HSCL‐37 comprised depression, anxiety, and somatisation items (Bean et al., 2007). Another widely used version with 25 items (HSCL‐25) comprised 10 anxiety and 15 depression items (Hesbacher, Rickels, Morris, Newman, & Rosenfeld, 1980) and was used as an outcome measure for Western inpatients (Derogatis, Lipman, Rickels, Uhlenhuth, & Covi, 1974). A review of the literature indicated that researchers and clinicians predominantly used HSCL‐25 to screen for depression and anxiety. Investigation of item factor structure has revealed two (depression and anxiety) to three (depression, anxiety, and somatisation) factors (Skogen, Øverland, Smith, & Aarø, 2017). HSCL‐25 has been used with refugee (Lavik, Laake, Hauff, & Solberg, 1999) and asylum‐seeker populations (Jakobsen, Thoresen, & Johansen, 2011) to screen for anxiety and depression symptoms. Researchers have also translated and adapted it to specific languages so that it can be used with vulnerable populations from refugee backgrounds (Baird & Skariah, 2016). Despite the use with CALD populations, and attempts at translations, very few have examined the factor structure and psychometric properties in CALD youth.

A derivative of the HSCL‐25 with 30 items (Hopkins Symptom Checklist‐30, HSCL‐30) has been used with Australian CALD adults (Khawaja, 2007). In a more recent study with CALD youth in Australia (Quinlan et al., 2016), an item on “loss of sexual interest and pleasure” was deleted to improve cultural sensitivity of the measure, consistent with previous research (Baird & Skariah, 2016). In addition, six somatisation items from the HSCL‐37 were added to improve cultural appropriateness of the scale (Hoge, Auchterlonie, & Milliken, 2006). In an Australian sample of Muslim migrants, the 30‐item version demonstrated good internal consistency and yielded a one‐factor solution (Khawaja, 2007). Similarly, the 29‐item scale demonstrated good internal consistency in a sample of CALD youth in Australia (Quinlan et al., 2016). Further Australian studies using these items with CALD youth (Khawaja, Allan, & Schweitzer, 2017; Khawaja, Ibrahim, & Schweitzer, 2017) provided provisional support for its reliability and validity.

The HSCL is widely used to screen for psychological distress across various settings and cultures but it has not been validated in CALD youth in Australia. A validated scale would have particular salience in the assessment of distress in school settings for the purpose of early intervention. The goal of the present study is to explore the factor structure and psychometric properties of the 29‐item HSCL in CALD youth in an Australian school setting. Since prior research indicated female gender was a risk factor for internalising psychological distress, the present study hypothesised that psychological distress would be higher in girls compared to boys (H1). Further, it was predicted that psychological distress would be higher in older participants, compared to that of younger participants (H2). Based on previous mixed findings where region of birth was a risk factor for psychological distress symptoms, a significant difference of ethnic origin was predicted but no direction hypothesised (H3). It was further hypothesised that children from refugee backgrounds would report greater symptomatology compared to children from migrant backgrounds (H4).

METHOD

Setting

The study was conducted at a specialised school in metropolitan Brisbane, which has been established to address the initial needs of children from refugee and migrant backgrounds. This specialised high school is a transitional school for CALD students to learn English. Youth attending the school ranged in age from 11 to 18-years old. Youth attend the school for 6 to 18-months, and are from the Middle East, Asia, Europe, Africa, and South America. The school offers a range of programs to enhance the students' adjustment and acculturation to Australia.

Participants

Two hundred and forty‐one participants, comprising 122 females (50.6%) and 119 males (49.4%) ranging in age from 11 to 18-years (M = 14.89, SD = 1.72) participated in the study. Of the sample, 52.3% were from refugee backgrounds, 41.5% were from migrant backgrounds, and 6.2% were unspecified. All participants were born overseas (46 different countries) and spoke a language other than English (46 different languages). Countries were merged into the following regions: Asia (49.4%), Africa (30.3%), South America (2.5%), Middle East (14.5%), and Europe (1.7%). English language proficiency varied with 13.4% being fluent in English, while 72.4% spoke English with some difficulty (11.7 experienced great difficulty) and 2.5% had no English‐speaking skills.

Measures

Demographic information

Participants' gender, age, country of birth, visa status (refugee, migrant), and English proficiency were collected on a demographic questionnaire.

Hopkins Symptom Checklist‐30

The HSCL‐30 is a variant of the HSCL‐25, with the addition of five items representing somatisation. For the present study 29 items were used and one item “loss of sexual interest and pleasure” was deleted to take into account the cultural considerations (Derogatis et al., 1974; Hesbacher et al., 1980; Khawaja, 2007; Quinlan et al., 2016). The scale comprises depression (14‐items), anxiety (10‐items), and somatisation (5‐items) subscales. Each item is scored on a 4‐point Likert scale ranging from 0 (not at all) to 4 (extremely). Higher scores indicate an increase in psychological distress. In line with previous research and cultural considerations, one item, “I am feeling blue” was modified to “feeling sad” and one item, “Loss of sexual interest or pleasure” was deleted (Baird & Skariah, 2016; Mels, Derluyn, Broekaert, & Rosseel, 2010). Previous studies report good internal consistency for the total scale in Australian CALD adults (α = .95; Khawaja, 2007) and for subscales in youth samples (anxiety α = .82, depression α = .88, somatisation α = .83; Quinlan et al., 2016).

Strengths and Difficulties Questionnaire‐Teacher Report (SDQ)

The 25‐item scale measures teachers' perceptions of students' psychological characteristics. It consists of five domains: conduct problems, emotional symptoms, hyperactivity, peer relationships, and prosocial behaviour (Goodman, 1997). The emotional symptoms subscale measures internalising symptoms, whereas the other four measure externalising symptoms (Grenier, Persson, & Rousseau, 2014). Items are measured on a 3‐point Likert scale ranging from 1 (not true at all) to 3 (certainly true). A total score can be calculated for all items and for each subscale. For each subscale, quantitative scale scores are categorised into normal (0–15), borderline (16–19), and abnormal (20–40) ranges. Previous research indicates sound internal consistency for the teacher report (α ≥ .86; Grenier et al., 2014).

Multidimensional acculturation stress scale

The 24‐item multidimensional acculturation stress scale measures acculturation stress. It comprises five dimensions: language barrier, homesickness, discrimination, threat to ethnic identity, and lack of opportunities for financial and occupational mobility (Jibeen & Khalid, 2010). In a study of Pakistani migrant adults in Canada, its internal consistency was reported satisfactory (α = .86; Jibeen & Khalid, 2010). For the present study, 14 items were selected (six from discrimination, four from homesickness, two from threat to ethnic identity and two from language barrier) and wording of “Canadians” was changed to “Australians” for items 14 to 18. The wording “Pakistani” was replaced by general phrases, for example, “From another country”. A higher score indicates a higher level of acculturative stress.

The Stirling Children's well‐being scale

The 15‐item Stirling Children's wellbeing scale (SCWBS) measures well‐being of youth aged between 8 and 15-years (Liddle & Carter, 2015). This positively worded scale was designed as an unidimensional measure. The authors found the scale to have good internal consistency (α ≥ .84), good construct validity and good test–retest reliability (Liddle & Carter, 2015). It has been used with CALD youth in Australia. These studies found the SCWBS as internally consistent (.89) (Khawaja et al., 2017). A higher score indicates a higher level of wellbeing.

School Connectedness

School connectedness was appraised by the 18‐item Psychological Sense of School Membership (PSSM) Scale (Goodenow, 1993). A 5‐point Likert scale (1 = Not at all to 5 = Completely) to indicate the degree to which respondents agreed with each statement about their various experiences of belonging to their school. The PSSM has been reported to have a Cronbach's alpha ranging from .77 to .82 (Goodenow, 1993). Analysis based on the Australian CALD youth indicated the Cronbach's alpha for the PSSM to be .87 (Khawaja et al., 2017).

Procedure

The current study was undertaken as part of a larger research project (Khawaja et al., 2017). Ethics approval was obtained from the university Ethics Committee. Informed consent for the youth was provided by the parent/guardian. Those who brought back signed consent forms were included in the study. Data were collected by a team of postgraduate students. These researchers were trained in administering the battery and working with CALD populations, with an emphasis on assessment on individuals with limited English proficiency. Subsequent to sufficient competency to deliver cross‐cultural assessment, researchers attended the School. Researchers spent time in the classrooms with the participants and helped the class teachers with various activities. This was conducted to increase participants' familiarity with the researchers. Due to the participants' limited English language proficiency, telephone interpreters were used to collect data. To reduce interpreter bias, prior to the data collection session, the researcher briefed the interpreter about the purpose of the study and the measures used to collect data. With the help of the interpreter, the purpose of the study was again explained to the participants and a verbal assent was obtained along with the previous written consent. Researchers explained confidentiality and participants' rights to withdraw from the study at any point without any impact on their relationship with the school. A verbal interview format recommended by previous researchers was used to gather information (Lhewa, Banu, Rosenfeld, & Keller, 2007). The questions, read out by the researcher, were translated into the participant's native language by the interpreter. The interpreter then conveyed the participant's responses to the researcher. Visual cues were used to help participants understand the Likert forms. The participant and the interpreter were debriefed after the session.

RESULTS

Data screening and assumptions

Nine participants, who did not complete HSCL were excluded from the data, while others with less than 10% missing data were retained. Values missing on seven HSCL items were deemed missing at random as shown by Little's MCAR test (χ2 [df = 200] = 199.18, p = .05). Missing data was estimated using the Expectation Maximisation method.

The dataset was screened for outliers and assumptions of normality were tested (Kline, 2013). Nine univariate outliers were found on HSCL but Mahalanobis distance indicated that they were not multivariate outliers. All HSCL items were positively skewed and 19 items exceeded kurtosis cut‐off (1.96). No variables were transformed as EFA (Exploratory Factor Analysis) is based on analysing the variances and covariances of variables rather than means. There was no evidence of multicollinearity. Bartlett's test of sphericity met the assumption of reliable correlations. Finally, sample size was deemed sufficient based on ratio of sample to variables of 5:1 (Tabachnick, Fidell, & Osterlind, 2001).

Exploratory factor analysis

Twenty‐nine HSCL items were subjected to EFA with principal axis extraction. As anxiety, depression, and somatisation are related constructs, the factors were expected to be non‐orthogonal. Thus, an oblique rotation method (direct oblimin) was employed. High sampling adequacy (Kaiser–Meyer–Olkin measure = 0.92) and sphericity (Barlett's Test = 3,211.97, df = 435, p < .001) of the correlation matrix indicated suitability for the factor model. EFA revealed seven factors that had eigenvalues greater than 1. Visual inspection of the scree plot indicated one factor based on the discontinuity principle (Cattell, 1966). Parallel analysis confirmed that only one factor met the criteria (O'Connor, 2000). Based on conceptual considerations and statistical criteria, one‐factor, two‐factor, and three‐factor solutions were considered. Principal axis extraction revealed that the one‐factor solution was the best fit based on statistical criteria and theoretical understandings. Table 1 presents the factor solution based on the 29 items.

Table 1. Factor loadings and communalities for one‐factor solution with 29 items

	Original item	Factor loading	Communality
Distress	1. Suddenly feel scared for no reason	0.53	0.28
	2. Feel faintness, dizziness, or weakness	0.53	0.28
	3. Feel nervousness or shakiness inside	0.61	0.37
	4. Heart pounding or racing	0.53	0.28
	5. Shake or tremble	0.68	0.46
	6. Feel tense	0.53	0.28
	7. Headaches	0.60	0.36
	8. Spells of terror or panic	0.53	0.28
	9. Feel restless, cannot sit still	0.65	0.42
	11. Blame yourself for things	0.54	0.29
	12. Cry easily	0.60	0.35
	13. Poor appetite	0.57	0.32
	14. Difficulty falling asleep, staying asleep	0.57	0.32
	15. Feel hopeless about the future	0.48	0.23
	16. Feel lonely	0.49	0.24
	17. Thoughts of harming yourself	0.51	0.26
	18. Thoughts of being trapped or caught	0.61	0.38
	19. Worry too much about things	0.68	0.46
	20. Feel no interest in things	0.62	0.39
	21. Feel everything is an effort	0.52	0.27
	22. Feel not good enough	0.53	0.28
	23. Feel critical of others	0.60	0.36
	24. Sweat	0.48	0.24
	25. Have to avoid certain things, places, activities because they frighten you	0.61	0.38
	26. Feel that your mind goes blank	0.58	0.34
	27. Have trouble concentrating	0.52	0.27
	28. Feel weak in parts of your body	0.66	0.44
	29. Feel sad	0.74	0.55
	30. Have heavy feelings in your arms and legs	0.71	0.51

Communalities and factor loadings of the 29‐item solution were examined to delete items with communalities below .30 and factor loading below .40. Although factor loadings of all items were above .4, communities of 13 items were below the cut‐off. An examination of these items indicated that the content of these items was abstract and difficult for the participants to comprehend (e.g., Feel hopeless about future) or perceived as irrelevant to distress (e.g., sweat). Therefore, these items were excluded from the analysis (Tabachnick et al., 2001). The 16‐item scale accounted for 40% of the variance. Table 2 displays the final factor loadings and communality estimates of the 16‐item unidimensional scale. The scale is referred as HSCL‐Y.

Table 2. Factor loadings and communalities for one‐factor solution with 16 items

	Original item	Factor loading	Communality
	3. Feel nervousness or shakiness inside	0.56	0.49
	5. Shake or tremble	0.68	0.57
	7. Headaches	0.56	0.37
	9. Feel restless, cannot sit still	0.66	0.50
	12. Cry easily	0.61	0.43
	13. Poor appetite	0.58	0.44
	14. Difficulty falling asleep, staying asleep	0.57	0.41
	18. Thoughts of being trapped or caught	0.61	0.41
	19. Worry too much about things	0.67	0.54
	20. Feel no interest in things	0.64	0.47
	23. Feel critical of others	0.58	0.41
	25. Have to avoid certain things, places, activities because they frighten you	0.63	0.42
	26. Feel that your mind goes blank	0.59	0.40
	28. Feel weak in parts of your body	0.68	0.48
	29. Feel sad	0.75	0.61
	30. Have heavy feelings in your arms and legs	0.74	0.59

Characteristics of the HSCL‐Y

The overall average score for total HSCL‐Y was 23.18 (SD = 8.47; range: 16–59).

Internal consistency for HSCL‐ Y full‐item scale was found to be excellent (α s= .91).

Convergent validity

Pearson's product moment correlations were calculated to compare the HSCL‐Y to the 29‐item HSCL, Strengths and difficulties questionnaire (SDQ) and Acculturation Stress. HSCL‐Y correlation with the 29‐item version, SDQ and Acculturative Stress was .97, .33 and .41 (p < .01) respectively. It seems that the shorter version is capturing the longer version effectively. Similarly, it is positively correlated with SDQ, the emotional symptoms of the participants observed and reported by their teachers, and acculturative stress. Subsequently, psychological distress reported by HSCL‐Y was associated with the distress identified by other measures.

Divergent validity

Pearson's product moment correlations revealed a significant negative correlation between HSCL‐Y and Well‐being (r = .41, p < .01) and School Connectedness (r = .32, p < .01). It seems that experiences of psychological distress, as captured by HSCL‐Y were contrary to positive experiences of well‐being and a sense of connectedness with the school.

Discriminant validity

Using the SDQ, the sample was divided into three groups; normal, borderline, and abnormal as is recommended (Goodman, 1997; Grenier et al., 2014). The total score of the HSCL‐Y were compared using analysis of variance (ANOVA). Between‐subjects ANOVA revealed a significant difference on distress for the normal, borderline and abnormal group on the SDQ emotional symptoms subscale, (F(2,179) = 1,190.00, p < .001). Bonferroni follow‐up comparisons revealed a significant difference between normal (M = 1.33, SD = 0.35) and abnormal groups (M = 1.80, SD = 0.73). No significant difference was found between normal and borderline and between borderline and abnormal.

Demographic differences

Between‐subjects ANOVAs were conducted to test demographic differences in reporting of psychological distress in the sample. The means of psychological distress were compared across region of birth (Asia, Middle East, Africa, Europe, and South America). A significant difference was found on distress between region of birth, (F(46, 193) = 1.61, p = .02). Follow‐up paired comparisons (REGWQ) revealed that youth from the Middle East and Asia were significantly higher on distress, respectively, compared to youth from Africa, Europe, and South America. Females scored higher (M = 24.24, SD = 9.56) than the males (M = 2.95; SD = 6.93). A significant difference was found on the reporting of distress between males and females (F(1,238) = 4.48, p = .05). No significant difference was found on the reporting of psychological distress for visa status (refugee or migrant), (F(1,224) = 0.06, p = .81); age (F(7,233) = 0.41, p = .90); nor English proficiency (F (3,235) = 1.11, p = .35].

DISCUSSION

With international migration, there is an increase in the resettlement of CALD youth in Western countries. To better respond to the mental health needs of CALD youth, there is a need to identify valid and reliable screening measures to assess CALD youth who may be psychologically distressed and may benefit from appropriate interventions. The culturally adapted version of HSCL with 29 items has been used in Australia with CALD youth. The present study undertook an exploratory approach and reports on a 16‐item unidimensional HSCL for this population demonstrating satisfactory internal consistency and validity. This scale identified differences in psychological distress based on gender and region of origin.

HSCL‐Y

Although various derivatives of HSCL have been reported in the literature, the 25‐item scale with the emphasis on anxiety and depression has been used extensively. Researchers in Australia modified the 25‐item version to a 30‐item scale to increase its cultural appropriateness, which was further reduced to 29 items to make it appropriate for CALD youth. Findings indicated that out of the 29 items, 16 items emerged as a single‐factor scale. The unidimensional structure is consistent with the previous finding based on adult Muslim migrants in Australia (Khawaja, 2007). The retained 16 items are conceptually clearer and probably easier to understand than the original scale. It is interesting to note that the items with lower communality were more abstract or ambiguous (e.g., feeling tense) or due to cultural differences not endorsed as a sign of psychological distress (e.g., sweating) by the CALD youth (Vonnahme et al., 2015). This finding is consistent with previous research, according to which HSCL was originally developed for a Western population; therefore it is unclear if the items measure the constructs in other populations in a similar manner (Syed et al., 2008).

These items reflect depression, anxiety, and somatisation as a single global factor. It is plausible that in this population, distress is a diffused experience, without distinctions between “depression,” “anxiety,” or “somatisation.” This is consistent with previous studies, which have noted the diffused nature of psychological distress in non‐Western populations (Terheggen, Stroebe, & Kleber, 2001; Miller, Madhur, & Kushner, 2006). The prevalent view of mind body holism and physical manifestations of psychological distress in non‐Western populations may account for the unidimensional construct of the HSCL in non‐Western samples. As seen in the literature, anxiety and depression are rarely found as separate dimensions in CALD samples (Scheper‐Hughes & Lock, 1987). Thus, the HSCL‐Y may thus be useful in identifying elevated levels of global distress.

Psychometric properties

The shorter 16‐item version captures the longer version of the HSCL‐Y effectively. Despite reduction in items the abbreviated scale is highly correlated with the 29‐item version. Consistent with previous studies utilising the 29‐item version with CALD youth (Quinlan et al., 2016), the 16 items are internally consistent and correlated with each other. The outcome is in line with the prior studies using various other versions to measure the same constructs (Kaaya et al., 2002; Lee et al., 2008; Syed et al., 2008). Further, HSCL‐Y has mild to moderate correlations with scale measuring acculturative stress and emotional distress. This outcome supports that HSCL‐Y's concurrent validity. Moreover, it emerged as conceptually different from School connectedness and Well‐being scales. Its negative correlations with these measures appraising positive experiences support its divergent validity. Finally, the discriminant validity was also supported and HSCL‐Y successfully differentiated a sample with higher emotional reaction from those who did not endorse any emotional experience (normal) and participants observed as emotionally distressed from those who were free from these symptoms. This suggests that the HSCL‐Y can be used as a screening tool in this sample and can be helpful in identifying psychological distress.

HSCL‐Y and demographic factors

Compared to that of the males, the level of psychological distress was significantly higher in females. This outcome is consistent with trends observed previously with CALD populations (Fazel et al., 2012; Renner & Salem, 2009). Moreover, it is interesting to note that differences in psychological distress emerged on the basis of participants' geographical region of origin. Those from Middle East and Asian origin reported a significantly higher distress than those from African, European, and South Americans regions. This finding is consistent with previous studies that found children from the Middle East were higher on depression scores compared to children from African and European regions (Bean et al., 2007; Hodes et al., 2008). In line with past studies it is possible that there could be more pre‐, peri‐, or post‐migration challenges for these subgroups (Hocking et al., 2015; Nickerson et al., 2015; Pacione et al., 2013). Further, CALD youth's distress could be related to the nature of the trauma they experience or cultural perception/manifestation. This is a new finding and requires further investigation. Nevertheless, this finding is contrary to the expectation that there would be no differences in psychological distress based on age or visa status. It is possible that participants were from a narrow age range and encountered similar experiences. Further, no difference between participants from refugee or migrant background indicates that despite the different migration pathways, participants' current experiences are probably similar. The interplay of risk factors for psychological distress in CALD youth is complex and difficult to differentiate.

Limitations and future directions

Although HSCL‐Y scale is shown to be valid and reliable, the study is not free from limitations. The test–retest reliability was not assessed. The scales stability over time requires investigation. It is important to examine whether the unidimensional factor is upheld. Future studies should use confirmatory factor analysis with data collected from diverse populations across Australia. The construct validity of the instrument needs to be assessed in a CALD population suggesting the need for further psychometric evaluation. Further, sample norms need to be developed including a baseline and clinical cut‐off for CALD youth. The data were collected from one school comprising a specific population of CALD children from refugee and asylum seeker and migrant backgrounds suggesting that the generalisability of the study is limited. Children from second generation ethnic minority populations are not represented. Considering the evidence that CALD populations are inhibited and avoid sharing their emotional experiences, it is unclear if the responses were impacted by social desirability (Whitley et al., 2006; Wu et al., 2009). Due to the language barriers, the data were collected with the help of telephone interpreters. Though some basic training was provided, it is not known if they translated the concepts correctly. Further, it is unknown whether meanings were lost in the interpretation process. It is possible that the conceptual and semantic differences in the interpretation of questions impacted on the validity of the scale (Tempany, 2009). To address this problem it may be helpful to replicate the study by using a few well‐trained interpreters consistently for the entire data collection. Further, the scale can be translated and back translated in other languages. Finally, HSCL was developed to measure psychological distress among the Western population. There is no evidence whether the construct is interpreted and perceived similarly by other non‐Western cultures. Further cross‐cultural investigation is warranted on the concept of psychological distress.

CONCLUSION

The findings report on the factor structure and psychometric properties of the HSCL‐Y in a sample of CALD youth. The results indicate that a 16‐item shorter version labelled as HSCL‐Y is a reliable and a valid measure that can be used to detect the psychological distress of CALD youth from refugee and migrant backgrounds. The findings are consistent with past studies supporting its use with CALD populations (Baird & Skariah, 2016; Jakobsen et al., 2011). Taking into account the importance of using screening measures in the school setting, HSCL‐Y may be a useful instrument considering that there are at‐risk children in schools who need to be identified quickly. Early detection can assist school authorities in directing the youth toward effective school and/or community‐based interventions (Beehler et al., 2012; Mitchelson et al., 2010). Further research is warranted to follow‐up cultural and theoretical considerations for a more confident use of this scale.

ACKNOWLEDGEMENTS

The authors would like to thank the participants and the school authorities from where the data were collected for their assistance.