Construct validity of the HSCL-25 and SCL-90-Somatization scales among Russian, Somali and Kurdish origin migrants in Finland

ABSTRACT

One of the most fundamental issues in psychiatric assessment across different sociocultural and linguistic groups is assuring test reliability and validity. We analyze the construct validity of the HSCL-25 depression and anxiety scales and the SCL-90-Somatization scale in a population-based study (Finnish Migrant Health and Wellbeing Study) among 1356 Russian, Somali and Kurdish origin migrants (aged 18–64 years old) in Finland using an Exploratory Structural Equation Modelling approach. The results revealed problems in establishing measurement invariance of the HSCL-25 depression and anxiety and SCL-90-Somatization scales. Two types of exploratory analysis (Exploratory Factor Analysis and Network Analysis) confirmed the mismatch between the theoretical structure and the data set. In addition, there were considerable group differences between the Russian, Somali and Kurdish groups, with the Somali group in particular showing differing results. The results are discussed in relation to sociocultural variation in manifesting psychological distress and psychiatric assessment in multicultural settings.

KEYWORDS:

• depression
• anxiety
• somatization
• migrants
• refugees
• test validity
• measurement invariance

Notes on contributors

Saija Kuittinen is a psychologist (MSc), philologist (MA), and a doctoral candidate at the University of Tampere, Finland. Her thesis focuses on sociocultural aspects of depression among Somali origin migrants in Finland.

Regina García Velázquez is a psychologist (MSc) and a doctoral student at the University of Helsinki, Finland. Her work focuses on psychometrics and on methodology applied to psychological assessment.

Anu E. Castaneda is a doctor of psychology, psychotherapist, and an expert of multicultural issues. As a research manager she leads the team of multicultural well-being in the National Institute for Health and Welfare (THL), Finland.

Raija-Leena Punamäki, Ph.D., is a psychologist and professor at the University of Tampere, Finland. Her research has focused on child development and mental health in conditions of war and military violence, psychosocial interventions for war-affected children as well as rehabilitation of victims of torture and human right abuse. Her current research focuses developmental aspects of trauma impacts, including infancy and adolescence, as well as family dynamics.

Shadia Rask (MSc) is a researcher and Ph.D. student in the team of multicultural well-being in the National Institute for Health and Welfare (THL), Finland. The focus of her thesis is psychological and physical functioning in migrant populations.

Jaana Suvisaari is research professor and head of Mental Health unit at the National Institute for Health and Welfare (THL), Finland. Her research has two focuses: research on psychotic disorders, spanning from etiological and pathophysiological mechanisms to early detection, somatic comorbidities and outcome, and population-based, longitudinal research on mental disorders and their risk factors and trajectories in the general population. She leads the mental health assessment in several THL’s general population surveys and is the PI of several clinical and epidemiological studies on psychotic disorders.

Notes

1. We use the term ‘sociocultural’ throughout this article because both social and cultural determinants are crucial in understanding mental health and wellbeing in different populations. However, it is often challenging to tease out cultural factors from social characteristics of a particular group. For example, besides cultural background, refugee origin groups often differ from the general population in factors related to education, employment, financial situation and social networks. For a discussion of the term ‘culture’ in comparative psychological research, see Poortinga (2015).

2. In addition, achieving full construct equivalence using solely statistical methods is challenging because some aspects that can be relevant for the construct (e.g., anxiety) in some group may be absent in the questionnaire.

3. In statistical terms they correspond to the concepts of structural and metric equivalence.

4. The majority of Finnish Kurds are from Iraq and Iran.

5. The health examination comprised various measurements of health, such as blood pressure, blood analysis, height, weight, functional capacity, oral health, a symptom interview and the HSCL-25 and SCL-90-S questionnaires. Most participants responded to the questionnaires themselves. Five Russian and 27 Kurdish speakers were interviewed by the research nurse because of the participants’ problems in answering the questions in the written format. In addition, 22 Russian, three Somali and three Kurdish participants requested help for answering from the research nurse.

6. Because categorical indicators were used (answer options were between 1 and 4), the estimation method was weighted least squares with mean and variance adjustment (WLSMV) with oblique target rotation. Theta parametrization was implemented.

7. Suicide and suicidal thoughts can be uncommon in some Muslim-faith populations because many interpret that the Koran forbids taking one’s own life (Koran 4:29). Of the Somali participants, 98% endorsed the answer option Not at all (see Table 5).

8. EFA was implemented based on the sample-weighted polychoric correlation matrices by Principal Axis extractions method because it does not make any assumptions about the underlying distribution of the data.

9. NA was calculated based on weighted partial polychoric correlations in order to control for redundancy of associations among variables.

10. The majority (83%) of the Kurdish participants reported speaking Sorani dialect of the Kurdish language as their mother tongue, and 10% reported speaking another Kurdish language dialect. Most of them identified themselves as ‘Kurds’ (63%) and a few as ‘Iranian’ (8%) or ‘Iraqi’ (3%).

11. Procrustes rotation compares the similarity between a matrix of loadings and a target matrix, indicated by the Tucker’s congruence coefficient. In order to run these calculations, we had to exclude the item Suicide because of its lack of variance in the Somali group.

Additional information

Funding

This work was supported by the European Social Fund (ESF).