scholarly journals Finding clusters of groups with measurement invariance: Unraveling intercept non-invariance with mixture multigroup factor analysis

2020 ◽  
Author(s):  
Kim De Roover
Keyword(s):  
Factor Analysis ◽  
Measurement Invariance ◽  
Behavioral Research ◽  
Invariant Factor ◽  
Factor Loadings ◽  
Latent Constructs ◽  
Daunting Task ◽  
Scalar Invariance ◽  
Metric Invariance ◽  
Latent Means

Comparisons of latent constructs across groups are ubiquitous in behavioral research and, nowadays, often numerous groups are involved. Measurement invariance of the constructs across the groups is imperative for valid comparisons and can be tested by multigroup factor analysis. For many groups, metric invariance (invariant factor loadings) often holds, whereas scalar invariance (invariant intercepts) is rarely supported. Scalar invariance is a prerequisite for comparing latent means, however. One may inspect group-specific intercepts to pinpoint non-invariances, but this is a daunting task in case of many groups. This paper presents mixture multigroup factor analysis (MMG-FA) for clustering groups based on their intercepts. Clusters of groups with scalar invariance are obtained by imposing cluster-specific intercepts and invariant loadings whereas unique variances, factor means and factor (co)variances can differ between groups. Thus, MMG-FA ties down the number of intercepts to inspect and generates clusters of groups wherein latent means can be validly compared.

scholarly journals Finding clusters of groups with measurement invariance: Unraveling intercept non-invariance with mixture multigroup factor analysis

2020 ◽  
Author(s):  
Kim De Roover
Keyword(s):  
Factor Analysis ◽  
Measurement Invariance ◽  
Behavioral Research ◽  
Invariant Factor ◽  
Factor Loadings ◽  
Latent Constructs ◽  
Daunting Task ◽  
Scalar Invariance ◽  
Metric Invariance ◽  
Latent Means

Comparisons of latent constructs across groups are ubiquitous in behavioral research and, nowadays, often numerous groups are involved. Measurement invariance of the constructs across the groups is imperative for valid comparisons and can be tested by multigroup factor analysis. For many groups, metric invariance (invariant factor loadings) often holds, whereas scalar invariance (invariant intercepts) is rarely supported. Scalar invariance is a prerequisite for comparing latent means, however. One may inspect group-specific intercepts to pinpoint non-invariances, but this is a daunting task in case of many groups. This paper presents mixture multigroup factor analysis (MMG-FA) for clustering groups based on their intercepts. Clusters of groups with scalar invariance are obtained by imposing cluster-specific intercepts and invariant loadings whereas unique variances, factor means and factor (co)variances can differ between groups. Thus, MMG-FA ties down the number of intercepts to inspect and generates clusters of groups wherein latent means can be validly compared.

scholarly journals Use of multi-group confirmatory factor analysis in examining measurement invariance in counseling psychology research

2017 ◽  
Vol 6 (1) ◽  
pp. 75-82 ◽  
Author(s):  
Hui Xu ◽  
Terence J. G. Tracey
Keyword(s):  
Factor Analysis ◽  
Confirmatory Factor Analysis ◽  
Measurement Invariance ◽  
Counseling Psychology ◽  
Latent Constructs ◽  
Configural Invariance ◽  
Invariant Measurement ◽  
Statistical Strategy ◽  
Confirmatory Factor ◽  
Metric Invariance

The purpose of this article is to introduce the theoretical implications and analytic strategies of measurement invariance. The article is focused on three important invariance conditions, consisting of configural invariance, metric invariance, and scalar invariance. Configural invariance refers to a qualitatively invariant measurement pattern of latent constructs across groups. Metric invariance refers to a quantitatively invariant measurement model of latent constructs across groups. Scale invariance refers to invariant mean levels of latent constructs across groups. While each invariance condition depicts one aspect of the relation between latent constructs with manifest observations, a progressive statistical strategy of measurement invariance was introduced based on multi-group confirmatory factor analysis. The article also provided a case example illustrating how to apply and examine measurement invariance in counseling psychology, with detailed theoretical implications and analytic decision-makings in each step. Application of measurement invariance in measurement comparison across multiple groups (e.g., gender, developmental stages, and national boundaries) was discussed and recommended.

scholarly journals Measurement Invariance of the Career Futures Inventory–Revised Across General and Client Samples

2018 ◽  
Vol 27 (4) ◽  
pp. 711-725
Author(s):  
Chan Jeong Park ◽  
Patrick J. Rottinghaus ◽  
Ze Wang ◽  
Ti Zhang ◽  
Nikki A. Falk ◽  
...  
Keyword(s):  
Measurement Invariance ◽  
Scale Validation ◽  
Career Adaptability ◽  
Analysis Model ◽  
Factor Analysis Model ◽  
Student Sample ◽  
Configural Invariance ◽  
Validation Procedure ◽  
Scalar Invariance ◽  
Latent Means

Establishing measurement invariance has been emphasized as an important scale validation procedure for group comparisons. The 28-item Career Futures Inventory–Revised (CFI-R) is a widely used measure of career adaptability that has demonstrated initial validity with various samples. The purpose of the present study is to further examine the validity of the CFI-R by testing measurement invariance between a general university student sample and a client sample. First, a five-factor confirmatory factor analysis model was tested with each group. Then, measurement invariance tests were conducted through subsequently examining configural invariance, metric invariance, and scalar invariance. Test of invariance was achieved until partial scalar invariance, suggesting that the CFI-R is similarly applicable to both clinical and nonclinical samples. In addition, the comparisons of latent means between two groups revealed that clients showed significantly lower latent means than general students for four factors: Career Agency, Occupational Awareness, Support, and Work–Life Balance.

Measurement validity in cross-cultural comparative research

2008 ◽  
Vol 17 (3) ◽  
pp. 211-220 ◽  
Author(s):  
Martin Prince
Keyword(s):  
Factor Analysis ◽  
Construct Validity ◽  
Measurement Invariance ◽  
Comparative Research ◽  
Cross Cultural ◽  
Research Translation ◽  
Factor Loadings ◽  
Criterion Validation ◽  
Recent Developments ◽  
Measurement Validity

SummaryBackground – The purpose of this article is to review the procedures to establish measurement validity in crosscultural comparative research, including recent developments in the quantitative assessment of cross-cultural construct validity. Methods – A narrative review, illustrated by selected examples, of methods in four areas – formative conceptual research, translation and adaptation, criterion validity and construct validity. Results – Valid assessment across cultures requires qualitative research to investigate the cultural relevance of the construct, a careful translation and adaptation of a common measure, followed by pre-testing and cognitive interviews on the populations to be tested. Full criterion validation across diverse cultures may be a chimera given the difficulty in establishing a universally applicable ‘gold standard'. Quantitative analyses can, however, have a part to play in establishing construct validity across cultures. Scale internal consistency, inter-item and item-total correlations and test-retest reliability provide basic support for the viability of a measure in a new cultural setting. Exploratory factor analysis can be used to compare factors and factor loadings. The hypothesis of ‘measurement invariance’ across countries and cultures can be tested explicitly using confirmatory factor analysis (common underlying factors and factor loadings) and Rasch models (common hierarchality of items). Despite measurement invariance, threshold effects arising from cultural differences in norms, or expectations, or expressions of mental distress may still be a problem. Conclusions – There are few examples in the cross-cultural mental health literature of demonstrably valid culture-fair comparison. Much more, could, in principle, be done either to demonstrate measurement invariance, or to identify and explore sources of heterogeneity.Declaration of Interest: None.

Assessing Perceptions of Programming Education Among P-12 School Teachers and Principals: A Multigroup Invariance Analysis

2018 ◽  
Vol 37 (6) ◽  
pp. 718-729 ◽  
Author(s):  
Siu-Cheung Kong ◽  
Yi-Qing Wang
Keyword(s):  
Measurement Invariance ◽  
Three Dimensions ◽  
Comprehensive Understanding ◽  
School Teachers ◽  
Programming Education ◽  
Scalar Invariance ◽  
Metric Invariance

This study aims to establish a robust measurement to obtain a comprehensive understanding of perceptions of programming education (POPE) of teacher and principal groups. In this study, the POPE scale contains three dimensions: (a) understanding in programming (UP), (b) support for programming (SP), and (c) expectation of programming (EP). Self-reported questionnaires were administrated among 258 teachers and 229 principals. Multigroup analyses were used among the two groups. The results of measurement invariance tests show that configural and metric invariance are fully supported, and scalar invariance is partially supported, suggesting the factor structures, loadings, and most item intercepts of the POPE scale are equivalent across the groups examined. Implications of the study are discussed.

Assessing the Measurement Invariance of the Gratitude Questionnaire–5 in Chinese and American Adolescents

2021 ◽  
Vol 24 ◽  
Author(s):  
Yu Ling ◽  
Qin Yang ◽  
Yifang Zeng ◽  
E. Scott Huebner
Keyword(s):  
Factor Analysis ◽  
Confirmatory Factor Analysis ◽  
Measurement Invariance ◽  
Factor Structure ◽  
Factor Analyses ◽  
Confirmatory Factor Analyses ◽  
Configural Invariance ◽  
Confirmatory Factor ◽  
Subsequent Comparison ◽  
Latent Means

Abstract Given the possibility of cultural differences in the meaning and levels of gratitude among children, we evaluated the measurement invariance of the Gratitude Questionnaire–5 (GQ–5) and differences in latent means across adolescents from two distinct cultures, China and America. Data were obtained from 1,991 Chinese and 1,685 American adolescents. Confirmatory factor analysis and multigroup confirmatory factor analysis were performed to examine the factor structure and the measurement equivalence across Chinese and American adolescents. The Cronbach’s alpha and Item-total Correlations of the GQ–5 were also evaluated. Results of confirmatory factor analyses provided support for the expected one-factor structure. Also, a series of multi-group confirmatory factor analyses supported full configural invariance, full metric invariance, and partial scalar invariance between the two groups. Furthermore, the findings suggested that the GQ–5 is suitable for conducting mean level comparisons. The subsequent comparison of latent means revealed that the Chinese adolescents reported significantly lower gratitude than American adolescents.

scholarly journals What to do when scalar invariance fails: The extended alignment method for multi-group factor analysis comparison of latent means across many groups.

2018 ◽  
Vol 23 (3) ◽  
pp. 524-545 ◽  
Author(s):  
Herbert W. Marsh ◽  
Jiesi Guo ◽  
Philip D. Parker ◽  
Benjamin Nagengast ◽  
Tihomir Asparouhov ◽  
...  
Keyword(s):  
Factor Analysis ◽  
Alignment Method ◽  
Group Factor ◽  
Scalar Invariance ◽  
Latent Means

scholarly journals Investigation of Interrater Reliability in The Evaluation of Foreign Language Writing Skills With Multigroup Confirmatory Factor Analysis

2018 ◽  
Vol 7 (1) ◽  
pp. 30
Author(s):  
Emine Önen ◽  
Melike Kübra Taşdelen Yayvak
Keyword(s):  
Factor Analysis ◽  
Confirmatory Factor Analysis ◽  
Foreign Language ◽  
Measurement Invariance ◽  
Interrater Reliability ◽  
Writing Skills ◽  
Preparatory School ◽  
Confirmatory Factor ◽  
Language Writing ◽  
Scalar Invariance

In this study, it was aimed to examine the interrater reliability of the scoring of paragraph writing skills on foreign languages with the measurement invariance tests.  The study group consists of 267 students studying English at the Preparatory School at Gazi University. In the study, where students write a paragraph on the same topic, the paragraphs are rated separately by three different interrater using the same scoring key. The evidence for the validity measurements was collected with AFA and DFA while the evidence for the reliability measurements was collected by the Cronbach-alpha (α) coefficient. As a result of testing with Multi-Group Confirmatory Factor Analysis within the context of the measurement invariance of the interrater reliability, no evidence of full and partial scalar invariance can be obtained while evidence of formal configural and metric invariance is obtained. As a result, the lack of evidence of scalar invariance means that raters scoring the writing skills do not use the same initial level of performance. In this case, the invariant uniqueness and invariant factor variances could not be tested, and therefore no evidence of reliability between raters could be obtained.

Measurement Invariance of Three Brief Measures of Rumination in Young Adults With and Without a History of Self-Injury

2020 ◽  
pp. 1-8
Author(s):  
Kate E. Tonta ◽  
Penelope Hasking ◽  
Mark Boyes ◽  
Joel Howell ◽  
Peter McEvoy ◽  
...  
Keyword(s):  
Measurement Invariance ◽  
Partial Metric ◽  
Total N ◽  
Thought Style ◽  
Configural Invariance ◽  
Self Injury ◽  
Item Functioning ◽  
History Of ◽  
Scalar Invariance ◽  
Metric Invariance

Abstract. Rumination is central to understanding the onset and maintenance of non-suicidal self-injury. Yet, no study has evaluated whether reported differences in rumination between people with and without a history of self-injury represent genuine group differences. The present study reports an investigation into the measurement invariance of three common measures of rumination in university students with and without a history of self-injury (total N = 1,519). Results revealed configural invariance for the Ruminative Responses Scale (RRS), the Ruminative Thought Style Questionnaire (RTSQ), and the Repetitive Thinking Questionnaire (RTQ). Additionally, the RTSQ and RTQ supported metric invariance, while the RRS supported partial metric invariance. Further, the RTQ demonstrated partial scalar invariance while the RTSQ demonstrated full scalar invariance. The current findings suggest that observed differences using the RTSQ and RTQ reflect genuine differences in rumination between people with and without a history of self-injury, while researchers using the RRS are advised to account for differential item functioning.

Sign in / Sign up

Export Citation Format

Share Document