Latent variables should remain as such: Evidence from a Monte Carlo study

Rdz-Navarro, Karina

doi:10.1080/00221309.2019.1596064

Cited by 8 publications

(3 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We then calculated the correlations between all content-related factors as well as the method factor of the 1FRIM and the latent variable of the PHQ-2 as a validity criterion. We chose to do so because previous research demonstrated that calculating correlations with latent variables should preferably be done within the SEM framework and not by extracting factor scores (Rdz-Navarro, 2019). In addition, we calculated the associations of the scales’ sum scores with the PHQ-2 to allow for evaluation of the impact of using a structural equation framework on the relationships compared to using manifest scale scores.…”

Section: Methodsmentioning

confidence: 99%

From Bi-Dimensionality to Uni-Dimensionality in Self-Report Questionnaires

Schmalbach

Zenger

Michaelides

et al. 2021

European Journal of Psychological Assessment

View full text Add to dashboard Cite

Abstract. The common factor model – by far the most widely used model for factor analysis – assumes equal item intercepts across respondents. Due to idiosyncratic ways of understanding and answering items of a questionnaire, this assumption is often violated, leading to an underestimation of model fit. Maydeu-Olivares and Coffman (2006) suggested the introduction of a random intercept into the model to address this concern. The present study applies this method to six established instruments (measuring depression, procrastination, optimism, self-esteem, core self-evaluations, and self-regulation) with ambiguous factor structures, using data from representative general population samples. In testing and comparing three alternative factor models (one-factor model, two-factor model, and one-factor model with a random intercept) and analyzing differential correlational patterns with an external criterion, we empirically demonstrate the random intercept model’s merit, and clarify the factor structure for the above-mentioned questionnaires. In sum, we recommend the random intercept model for cases in which acquiescence is suspected to affect response behavior.

show abstract

Section: Methodsmentioning

confidence: 99%

From Bi-Dimensionality to Uni-Dimensionality in Self-Report Questionnaires

Schmalbach

Zenger

Michaelides

et al. 2021

European Journal of Psychological Assessment

View full text Add to dashboard Cite

show abstract

“…Instead, a single language factor provided the best fit to the data (χ 2 = 7.37, df = 8, p = .497, comparative fit index [CFI] = 1.00, Tucker-Lewis index [TLI] = 1.00, root mean square error of approximation [RMSEA] = .00, 90% CI [.00, .058]), with factor loadings ranging from .60 to .81 across all six measures. The one-factor model had a high reliability of omega of .90, while its factor scores also had a high reliability of .90 (online ), hence minimizing the possibility of attenuation of associations due to measurement error (Rdz-Navarro, 2019).…”

Section: Methodsmentioning

confidence: 99%

Sources of variability in the prospective relation of language to social, emotional, and behavior problem symptoms: Implications for developmental language disorder.

Yew¹,

Griffiths²,

Norbury³

et al. 2021

Journal of Abnormal Psychology

View full text Add to dashboard Cite

Children with developmental language disorder (DLD) are at risk for social, emotional, and behavioral (SEB) maladjustment throughout development, though it is unclear if poor language proficiency per se can account for this risk as associations between language and SEB appear more variable among typical-language children. This study investigated whether the relationship between language and SEB problems is stronger at very low levels of language and considered confounders including socioeconomic status, sex, and nonverbal intelligence. These were examined using a population-based survey design, including children with a wide range of language and cognitive profiles, and assessed using the Strengths and Difficulties Questionnaire and six standardized language measures (n = 363, weighted n = 6,451). Structural equation models adjusted for prior levels of SEB revealed that the relationship of language at age 5-6 years to SEB at 7-9 years was nonlinear. Language more strongly predicted all clusters of SEB at disordered language levels relative to typical language levels, with standardized betas of À.25 versus .03 for behavioral, À.31 versus À.04 for peer, and .27 versus .03 for prosocial problems. Wald tests between these pairs of betas yielded p values from .049 to .014. Sex moderated the nonlinear association between language and emotional symptoms. These findings indicate a clinical need to support language development in order to mitigate against problems of SEB and to carefully monitor the mental health needs of children with DLD, particularly in the context of multiple, and potentially sex-specific, risks. General Scientific SummaryChildren with a marked deficit in language have higher rates of behavioral and emotional disorders in later life. However, studies among typically developing children do not always find a link between language and symptoms of emotional behavioral disorders. This study supports the notion of unequal risk, where the risk is amplified for children with very low levels of language.

show abstract

“…While the asymptotic variance of WLE continues to be biased, its bias is smaller than that of MLE. As MLE is theoretically unbiased, so are WLE (Rdz-Navarro, 2019). In this study, we used a 2-PL generalized partial credit model (GPCM) for the responses.…”

Section: Weighted Maximum Likelihood Estimator (Wle) Scoresmentioning

confidence: 99%

To Score or Not to Score? A Simulation Study on the Performance of Test Scores, Plausible Values, and SEM, in Regression With Socio-Emotional Skill or Personality Scales as Predictors

Bhaktha

Lechner

2021

Front. Psychol.

View full text Add to dashboard Cite

This article addresses a fundamental question in the study of socio-emotional skills, personality traits, and related constructs: “To score or not to score?” When researchers use test scores or scale scores (i.e., fallible point estimates of a skill or trait) as predictors in multiple regression, measurement error in these scores tends to attenuate regression coefficients for the skill and inflate those of the covariates. Unlike for cognitive assessments, it is not fully established how severe this bias can be in socio-emotional skill assessments, that is, how well test scores recover the true regression coefficients — compared with methods designed to account for measurement error: structural equation modeling (SEM) and plausible values (PV). The different types of scores considered in this study are standardized mean scores (SMS), regression factor scores (RFS), empirical Bayes modal (EBM) score, weighted maximum likelihood estimates (WLE), and expected a posteriori (EAP) estimates. We present a simulation study in which we compared these approaches under conditions typical of socio-emotional skill and personality assessments. We examined the performance of five types of test scores, PV, and SEM with regard to two outcomes: (1) percent bias in regression coefficient of the skill in predicting an outcome; and (2) percent bias in the regression coefficient of a covariate. We varied the number of items, factor loadings/item discriminations, sample size, and relative strength of the relationship of the skill with the outcome. Results revealed that whereas different types of test scores were highly correlated with each other, the ensuing bias in regression coefficients varied considerably. The magnitude of bias was highest for WLE with short scales of low reliability. Bias when using SMS or WLE test scores was sometimes large enough to lead to erroneous research conclusions with potentially adverse implications for policy and practice (up to 55% for the regression coefficient of the skill and 20% for that of the covariate). EAP, EBM, and RFS performed better, producing only small bias in some conditions. Additional analyses showed that the performance of test scores also depended on whether standardized or unstandardized scores were used. Only PV and SEM performed well in all scenarios and emerged as the clearly superior options. We recommend that researchers use SEM, and preferably PV, in studies on the (incremental) predictive power of socio-emotional skills.

show abstract

Latent variables should remain as such: Evidence from a Monte Carlo study

Cited by 8 publications

References 37 publications

From Bi-Dimensionality to Uni-Dimensionality in Self-Report Questionnaires

From Bi-Dimensionality to Uni-Dimensionality in Self-Report Questionnaires

Sources of variability in the prospective relation of language to social, emotional, and behavior problem symptoms: Implications for developmental language disorder.

To Score or Not to Score? A Simulation Study on the Performance of Test Scores, Plausible Values, and SEM, in Regression With Socio-Emotional Skill or Personality Scales as Predictors

Contact Info

Product

Resources

About