Assessing the factor structure of the Spanish language parent Strengths and Difficulties Questionnaire (SDQ) in Honduras

Abstract

With worldwide prevalence rates between 10% and 20%, mental illness in children and adolescents is an issue for which culturally sensitive screening tools are needed. The Strengths & Difficulties Questionnaire (SDQ) is a commonly used measure that has been translated into numerous languages, although some research suggests issues with cross-cultural validity. Only four other studies have tested the Spanish-language SDQ in Latin America. In this study, we aimed to help fill this gap by assessing the factor structure of the parent or teacher version of the Spanish-language SDQ (for children ages 4–17) with 967 parent or other caregiver respondents of primary school-aged children (ages 4 to 17) in the Department of Intibucá, Honduras. When unable to find a good fitting factor model previously identified in the literature, we conducted split sample exploratory factor analysis (EFA) and confirmatory factor analysis (CFA), along with measurement invariance testing with the best fitting EFA-extracted model based on gender for caregiver respondent and child gender. Results showed that while many EFA models had a good fit, CFI and TLI was < .90 for all extracted models when confirmed in the second sample with CFA. We then modified the best fitting extracted three-factor, 24-item model, which dropped item 15, by allowing select item residuals to correlate, increasing CFI and TLI to > .90 for female child gender. This modified three-factor model was partially invariant for configural and scalar models between child genders. Configural and scalar models would not converge for adult genders. Of note, metric models were not produced in Mplus related to items’ cross-loading on more than one factor. These findings suggest that the SDQ Spanish language parent or teacher version may not perform optimally cross-culturally in this area of Honduras. Future research should confirm these findings with other Honduran populations.

Reverse-Coded Items Do Not Work in Spanish: Data From Four Samples Using Established Measures

“…However, for the Openness scale, excluding reverse-coded items improved scale consistency and, further, un-reversing and including items was only marginally different from including them per the scoring instructions (alpha of 0.632 versus 0.649; omega 0.654 versus 0.670). In the caregiver sample, the SDQ demonstrated low item-total correlations for reverse-coded items across the board and, in two instances, the correlation actually increased when the items were Brown et al, 2014;Harry et al, 2019]. The DERS demonstrated fewer problems with reverse-coded items.…”

“…Likewise, Cronbach’s alpha and McDonald’s omega revealed an improvement in scale functioning when reverse-coded items were excluded and remarkably little difference when including them in an un-reversed, incorrect manner (alpha of 0.686 versus 0.582; omega 0.271 versus 0.314). Still, internal consistency estimates (both Cronbach’s alpha and McDonald’s omega) were low, indicating problems of reliability and echoing prior research calling into question the internal consistency and factor structure of the SDQ when administered in Spanish [(Author self-citation), Brown et al, 2014 ; Harry et al, 2019 ]. The DERS demonstrated fewer problems with reverse-coded items.…”

“…Items are rated from “not true” = 0 to “certainly true” = 2. Versions of the SDQ have generally shown adequate reliability as well as good criterion and convergent validity ( Hill and Hughes, 2007 ; Tsang et al, 2012 ; Harry et al, 2019 ). Specifically, the Spanish caregiver version of the SDQ further demonstrates such strong efficacy ( Gómez-Beneyto et al, 2013 ; Ortuño-Sierra et al, 2018 ; Harry et al, 2019 ), with reports of sufficient internal consistency reliability (McDonald’s ω = 0.84, Ordinal α = 0.75–0.81) in addition to similar factor structure and construct validity across studies utilizing other caregiver SDQ versions ( Koskelainen et al, 2001 ; Percy et al, 2008 ; Ruchkin et al, 2008 ).…”

First data and translation of the Child Attachment Interview en Español.

“…Items are rated from not true = 0 to certainly true = 2. Versions of the SDQ scores have generally shown adequate reliability as well as good criterion and convergent validity (Harry et al, 2019; Hill & Hughes, 2007; Tsang et al, 2012), including the Spanish caregiver version (Gómez-Beneyto et al, 2013; Harry et al, 2019; Ortuño-Sierra et al, 2018). This instrument was used in the caregiver sample.…”

Validation of the Spanish version of the Pediatric Symptom Checklist (PSC) to identify and assess psychosocial problems among early adolescents in Chile

“…It has three versions: self-reported, parents/carers-reported, and teachers-reported. It contains five subscales with five items each subscale: emotional symptoms, conduct problems, hyperactivity/attention problems, peer problems, and prosocial behavior [ 27 ]. The first four subscales are part of the total difficulties scale, while the last subscale is considered to be a part of the strengths [ 19 ].…”