Factor Structure of the Ruminative Response Scale and Measurement Invariance across Gender and Age among Chinese Adolescents

Abstract

Methods: Participants included 1103 adolescents from Hunan Province and Hainan Province in China. A confirmatory factor analysis (CFA) was conducted to determine the structure of the RRS on a large scale. Multiple CFAs were utilized to test the measurement invariance of the three subscales included in the RRS, across gender and age. Results: The RRS exhibited strong internal consistency (overall Cronbach's alpha coefficient = 0.927) and internal reliability (0.362-0.864). The CFA showed a valid three-factor model. Measurement invariance of the model across gender and age was fully supported. Conclusions: Total RRS and subscales have accepted validity and reliability. Three-factor model of the RRS across gender and age was established. Measurement equivalence of the RRS across gender and age was established among Chinese adolescents. These findings suggest that the use of the RRS in comparing samples across gender and age is valid.

Examining the Measurement Invariance of TIMSS 2015 Mathematics Liking Scale through Different Methods

“…MG-CFA, a commonly preferred method in measurement invariance studies in various disciplines (Meredith, 1993;Mullen, 1995;Steenkamp & Baumgartner, 1998), is a parametric and linear approach investigating the similarity between measurement model parameters named as factor loadings, intercepts and error variances for the same factor models across groups. Measurement invariance within the scope of MG-CFA is defined and tested through four hierarchical models (Byrne & Stewart, 2006;Meredith, 1993;Wu, Li & Zumbo, 2007). The measurement invariance levels that are tested in MG-CFA can be listed respectively as follows:…”

“…In the conclusion part of their study, they stated that the measurement invariance failed to be established because of translation-related problems and cultural differences. Wu, Li, and Zumbo (2007) investigated the cross-country measurement invariance using TIMSS 1999 data in their study. Accordingly, by using the mathematics achievement scores of 21 countries participating in TIMSS 1999, it was checked whether the measurement invariance was achieved with MG-CFA.…”

“…The Mixed Rasch Model can be used in the analysis of the tests measuring the affective traits as well as in the achievement tests (Rost, Carstensen & von Davier, 1997). Many studies tested the measurement invariance by means of MRM (Aryadoust, 2015;Aryadoust & Zhang, 2016;Cohen & Bolt, 2005;Eid & Rauber, 2000;Pishghadam, Baghaei & Seyednozadi, 2017;Şen, 2016;Yalçın, 2019;Yüksel, 2015;Yüksel, Elhan, Gökmen, Küçükdeveci & Kutlay, 2018). Tee and Subramaniam (2018) analyzed the measurement invariance of the attitudes towards eighth grade science in the UK, Singapore and USA countries that entered TIMSS 2011 with Rasch analysis.…”

“…There are a number of studies in the literature conducting cross-cultural comparisons thanks to the accessibility to international data (e.g. Alatlı, Ayan, Demir & Uzun, 2016;Asil & Gelbal, 2012;Rutkowski & Svetina, 2014). In international assessments such as TIMSS, data is collected by administering a single measurement instrument to all participants from different countries.…”

“…These methods are combinations of a latent trait or the latent class analysis and a structural equation model (Eid & Rauber, 2000). Multiple Group Latent Class Analysis (MG-LCA), which is a method among mixed distribution models and is dependent on latent class analyses, may also be employed in measurement invariance studies (Magidson & Vermunt, 2001;Moors, 2004;Moors & Wennekers, 2003). MG-CFA is the method which is used when the observed and latent variables are continuous but cannot be used when both are categorical (Somer, Korkmaz, Dural & Can, 2009).…”