Internal consistency and structural validity of the parent-report preschool (2-4 years) Strengths and Difficulties Questionnaire in 1-year-old children

TL;DR

This study evaluated a questionnaire for assessing child mental health in 1-year-olds and found it unreliable and structurally invalid for this very young age group.

Contribution

The study is the first to examine the psychometric properties of the preschool Strengths and Difficulties Questionnaire in children younger than 2 years.

Findings

The two-factor model of the questionnaire showed poor fit (CFI = 0.612, RMSEA = 0.079) and low internal consistency for the internalising subscale (ω = 0.57).

The five-factor model also had poor fit (CFI = 0.676, RMSEA = 0.070) with three out of five subscales below acceptable internal consistency.

The study concludes that the questionnaire lacks validity for children younger than 2 years and requires adaptation for this age group.

Abstract

Prevention and early intervention are key to addressing poor child mental health. Systematic reviews have highlighted a lack of brief, valid and reliable outcome measures that can be implemented in both research and practice to assess social, emotional and behavioural outcomes in the early years. The Preschool Strengths and Difficulties Questionnaire (2–4 years) is a promising candidate to fill this gap, but the measurement properties of this tool are not yet known in very young children.

A secondary data analysis of two clinical trial datasets was conducted to examine the internal consistency reliability and structural validity of the parent-report English preschool version of the Strengths and Difficulties Questionnaire in a sample of 505 infants with mean average age of 18 months (SD .81). The measure was designed for children aged 2–4 years and was not modified prior to use with 1-year-olds in this study. Structural validity was examined in two Confirmatory Factor Analyses (CFA) testing two-factor and five-factor models (representing factor structures proposed by the developers of SDQ), and McDonald’s coefficient Omega was estimated for each subscale with values > .70 considered acceptable.

The model fit values for the two-factor model demonstrated a poor fit to the data (X2 = 626.067(151) = p < .001, CFI = 0.612, RMSEA = 0.079 [90% CI .073 to .085], SRMR = .077) and the omega value was below acceptable at ω = .57 for the internalising subscale and ω = .76 for the externalising subscale. The five-factor model also demonstrated a poor fit to the data (X2 = 836.813(242) = p < .001, CFI = 0.676, RMSEA = 0.070 [90% CI .065 to .075], SRMR = .081). Omega values were below acceptable for three out of five subscales.

We concluded that the measure has poor internal consistency and lacks structural validity in this very young age group. Further research to adapt the SDQ in order to improve content and face validity is recommended prior to any further psychometric analyses with this very young age group. The paucity of robust and practical outcome measures of early social, emotional and behavioural poses significant challenges to the early identification of need and evaluation of interventions.

The online version contains supplementary material available at 10.1186/s41687-025-00905-1.

Figures