A guide to exploratory structural equation modeling (ESEM) and bifactor-ESEM in body image research

The Intuitive Eating Scale-3: Development and psychometric evaluation

“…MIMIC tests of DIF were performed in the following sequence (Morin et al, 2013) In Studies 1-3, analyses for composite reliability and validity include both latent (in standardized units whereby M = 0, SD = 1) and observed (manifest) scores. While latent scores are more accurate than observed scores (i.e., averaging item responses; Swami, Maïano, & Morin, 2023), observed scores are more accessible to researchers and clinicians. Furthermore, nearly all research on intuitive eating has used observed scores and, therefore, we provide observed scores for an "apples-to-apples" comparison with previous research using observed IES-2 or IES scores.…”

“…Given that all IES-3 items are novel, we conducted exploratory factor analysis (EFA) as the first analytic step to determine the appropriate number of factors that should be retained without any a priori modeling limitations (that is, EFA allowed us to adopt a data-driven approach to determine how best to model our data without any theoretical constraints; Swami, Maïano, & Morin, 2023). We hypothesized that four factors-corresponding to the UPE, EPR, RHSC, and B-FCC domains-would yield good fit to the data (H1) and that factor scores would demonstrate adequate composite reliability (H2).…”

“…While the IES-2 had a proposed an overarching factor, Tylka and Kroon Van Diest (2013) established this via higher-order modeling. However, higher-order modeling is now recognized as a flawed approach, as it makes the unrealistic assumption that associations between items and the higher-order factor are entirely mediated by their respective lower-order factors (Swami, Maïano, & Morin, 2023). Instead, a global, bifactor approach provides a more viable alternative than higher-order models, as items are allowed to define a global (G-)factor as well as one specific (S-)factor (Swami, Maïano, & Morin, 2023).…”

“…However, higher-order modeling is now recognized as a flawed approach, as it makes the unrealistic assumption that associations between items and the higher-order factor are entirely mediated by their respective lower-order factors (Swami, Maïano, & Morin, 2023). Instead, a global, bifactor approach provides a more viable alternative than higher-order models, as items are allowed to define a global (G-)factor as well as one specific (S-)factor (Swami, Maïano, & Morin, 2023). Indeed, the available evidence vis-à-vis the IES-2 indicates poor fit of higher-order models and superior fit of bifactor models (e.g., Ramalho et al, 2022;Swami et al, 2022;Swami, Maïano, et al, 2021).…”

The Experience of Embodiment Scale: An examination of its psychometric properties in women from the Republic of Cyprus

“…Next, we also cross-validate the optimal model and consider multidimensional functioning (i.e., higher-order and bifactor models), which would provide the most rigorous test of the factorial validity of the EES to date. This analytic framework has been employed previously [ 27 ] and allows us to honour the theoretical foundations of the EES while ensuring that empirical considerations are not minimised [ 28 ].…”