Challenging the Status Quo: A Critical Examination of the Sibling Relationship Questionnaire's Construct Validity

Background: To date, over half (n=8, 61.5%) of the 13 widely-used measures of sibling relationship quality, including Furman & Buhurmester’s Sibling Relationship Questionnaire (SRQ), were developed between the 1980s and early 2000s using samples of white, two-parent families. Less than half (n=5, 38.5%) employed now-standard factor analytic techniques to define underlying latent constructs, likely influenced by constraints of early estimation techniques. Such limitations in statistical computing resources precluded application of more powerful psychometric approaches to analysis. Given the importance of incorporating sibling relationship quality operationalization into research on child psychosocial development, it is critically important to evaluate such measures’ construct validity. We examined the construct validity of the SRQ, one of the most commonly used measures of sibling relationship quality among elementary school-age children.

Methods: Data included parent-reported SRQ item scores for 476 socioeconomically diverse U.S. children (N = 238 sibling dyads) between the ages of 5-12 years collected in 2020. As initially described by Furman & Buhrmester, the 48 originally developed SRQ items comprise 16 subdimensions (e.g., companionship, dominance of sibling, antagonism, maternal partiality) that cluster into four broader dimensions of sibling relationship quality including warmth/closeness, relative status/power, conflict, and rivalry. Exploratory factor analytic (EFA) models using maximum likelihood factor extraction were estimated using Mplus to evaluate the degree to which the factor configuration (i.e., four primary dimensions of sibling relationship quality) described in the initial SRQ development could be empirically reproduced and thus evidence its construct validity.

Results:

EFA results did not evidence a clear factor structure of the SRQ, even when specifying up to four factors to mirror the commonly described four core domains of the measure (sibling warmth/closeness, relative status/power, conflict, and rivalry). Based on the Comparative Fit Index (CFI), model fit statistics of the unidimensional, two-, three-, and four-factor models were .46, .63, .73, and .78, respectively. After freeing the model to estimate up to seven underlying factors, the 7-factor (and comparatively best-fitting) model failed to adequately reproduce the data (CFI=.88, RMSEA=.06, RMSEA 95% CI .05-.07). Further examination of EFA findings suggests the presence of various methods factors (e.g., multiple items with identical item stems clustering together into a common factor) in addition to possible trait factors (i.e., substantive domains characterizing sibling relationship quality).

Conclusion: We failed to reproduce Furman & Buhrmester’s often-cited structure of sibling relationship quality when employing standard psychometric techniques,calling into question the construct validity of other sibling relationship quality measures similarly developed (e.g., without rigorous psychometric testing). Rigorous measurement development is critically needed in the field of sibling research. Measures must be developed using robust psychometric techniques and across diverse samples of siblings. Without these measures, understanding of this formative relationship in childhood is limited and hinders the development of effective interventions to leverage this unique social-emotional bond to nurture child well-being.