Leaning into AI for Qualitative Analysis: An Exploratory Methodological Study

Background: Siblings are one of the most foundational relationships in a child’s life and endure well beyond a child’s relationship with their parents. In the mid-1980s, researchers began to examine sibling relationship quality to understand what type of sibling relationship qualities predict child well-being outcomes. However, findings are mixed regarding the association between sibling relationship qualities and child outcomes. To our knowledge, no research has systematically examined the methodological considerations of sibling relationship quality measures used in these studies. We conducted a systematic review on general population sibling relationship quality among children aged 2 to 17 in the United States and examined each measure for utilization of robust instrument development and validation standards as defined by the NIH PROMIS.®

Methods: Six electronic bibliographic databases (PsychInfo, Medline, CINAHL, SocIndex, SSCI, Scopus) were searched for articles published up until February of 2022 using the following search terms: (sibling* or “sibling relation*” OR brother* OR sister*) AND (relationship N3 quality) AND (child* OR youth OR adolescen* OR toddler* OR infant*). After deduplication, two investigators coded each citation for inclusion if the study (a) examined sibling relationship quality; (b) sample included children between the ages of 2-17 years old; and (c) was published in a peer reviewed journal. Citations were excluded if (a) the study examined child response to a birth of a sibling, sibling incest, or sibling abuse; (b) study was conducted outside of the United States; (c) article was not published in English; and (d) sample included special populations such as foster care siblings, sibling with a developmental disability, or sibling with a severe health condition.

Results: The search yielded 947 citations after deduplication with a total of 87 articles on general population sibling relationship quality that utilized 13 measures of sibling relationship quality. Across all 13 measures, only one measure utilized a robust mixed-methods study design (i.e., qualitative analysis, EFA, CFA) for measurement development. However, none of the 13 measures identified through the systematic review met all criteria specified by PROMIS® for developing reliable and valid measures. The validity of existing measures is compromised due to serious limitations in their development such as narrow sampling (i.e., primarily white, two-parent, middle-to-upper class families), item development and refinement (i.e., minimal use of content expert consultation, interviews with children and families), and analytic approaches (i.e., reliance on reliability tests and principal components analytic techniques rather than those of exploratory or confirmatory factor analysis).

Conclusions: Underdeveloped measurement tools have hindered researchers and clinicians from fully understanding and developing related interventions that capitalize on the potential of sibling relationships. The lack of rigorous development processes across measures of sibling relationship quality impacts the reliability and validity of these measures, their appropriateness across different subpopulations of children, and compromises subsequent modeling with an imprecise measure of sibling relationship quality. No significant measurement work has been done since the 1980s/1990s, providing an opportunity to develop and test updated measures of sibling relationship quality that better reflect current family and cultural contexts.