Sensitivity analysis of structural equation model with non-normal observed variables

PurposeThe purpose of this paper is threefold: developing an effectiveness communication model for a city rotary event. Empirical validation of the new effectiveness model based on a city cultural festival event to explore the similarity and difference of perception change on event attendance. Providing a structural equation modeling (SEM) analysis of communication model for both professional event organizers and PCO guidelines when determining an event’s marketing strategy priorities.Design/methodology/approachThe authors use SEM to verify the relationship between variables, and path coefficient comparison analysis which was adapted for model verification and hypothesis testing. Analytical methods included reliability analysis, descriptive statistical analysis, HotellingT2-test, SEM analysis.FindingsThis result demonstrates that attendee increased their emotion, attitude, city brand image as result of the event activities, and positive event emotions have the highest influence toward to a positive city’s brand emotion. A positive city brand emotion especially has the second highest influence on attendee next event retentions. A positive city brand emotion will lead to a positive city brand attitude, but city brand attitudes will not have a positive impact on retentions.Originality/valueThe contribution and value of this study is that the authors established a new effectiveness communication model which introduce the city branding concept into the research. The study provides empirical evidence of the effectiveness illustrations of the application in the communication model. The research meets the objective in this effectiveness communication model. It illustrates the directions for how an event should be designed in order to create a positive city brand attitude for retention.

The association between adverse childhood experiences (ACEs) and depression has been the focus of a number of prevalent studies in recent years—particularly among high-risk youth. Depression remains a significant mental health issue among justice-involved youth. There is a well-established correlation between depressed mood and conduct problems (e.g., conduct disorder and oppositional defiant disorder) during childhood and adolescence, which tends to become more prevalent during adolescence. Studies of justice-involved youth reveal high prevalence rates of depression and other mood disorders. Drawing on the relevant literature, we conducted multigroup structural equation model (SEM) analyses to assess the relationships between experiencing ACEs, sexual assault victimization, and depression among male (n = 226) and female (n = 98) youth entering a post-arrest intake facility in the Florida, U.S.A. juvenile justice system in 2024–2025. The youths averaged 15 years in age, and most were attending middle school or high school. Confirmatory factor analyses (CFAs) were completed to estimate a latent variable labeled depression. Sexual assault victimization and ACEs were hypothesized to be related to each other and were specified as predictors of depression. This trauma/stress experiences and depression model was estimated in two multigroup analyses, across birth gender groups (male or female) and race groups (non-Black or Black) for the youth in this study. The results indicated that there are several notable conclusions from the SEM analyses. First, depression was a scalar invariant in the two multigroup analyses, permitting clearer comparisons of the specified predictors of this construct across groups. Second, for the race-based SEM, experience of sexual assault and the total ACE score were significantly related only in the model for Black youth. The fit of the model was “poorest” among non-Black youth, although even in this case, sexual assault experiences were a significant predictor of depression. Finally, for the gender-based model, sexual assault and ACEs were significant predictors of depression among both male and female youth. Model fit results underscore the important role of abuse trauma and ACEs in understanding these youths’ depression symptoms, and they help contribute to the literature on this topic.

This paper presents the usage of Structural Equation Modelling (SEM) approach in analysing the effects of Building Information Modelling (BIM) technology adoption in improving the reliability of cost estimates. Based on the questionnaire survey results, SEM analysis using SPSS-AMOS application examined the relationships between BIM-improved information and cost estimates reliability factors, leading to BIM technology adoption. Six hypotheses were established prior to SEM analysis employing two types of SEM models, namely the Confirmatory Factor Analysis (CFA) model and full structural model. The SEM models were then validated through the assessment on their uni-dimensionality, validity, reliability, and fitness index, in line with the hypotheses tested. The final SEM model fit measures are: P-value=0.000, RMSEA=0.079<0.08, GFI=0.824, CFI=0.962>0.90, TLI=0.956>0.90, NFI=0.935>0.90 and ChiSq/df=2.259; indicating that the overall index values achieved the required level of model fitness. The model supports all the hypotheses evaluated, confirming that all relationship exists amongst the constructs are positive and significant. Ultimately, the analysis verified that most of the respondents foresee better understanding of project input information through BIM visualization, its reliable database and coordinated data, in developing more reliable cost estimates. They also perceive to accelerate their cost estimating task through BIM adoption.

Meta-analysis plays a key role in combining studies to obtain more reliable results. In social, behavioral, and health sciences, measurement units are typically not well defined. More meaningful results can be obtained by standardizing the variables and via the analysis of the correlation matrix. Structural equation modeling (SEM) with the combined correlations, called meta-analytical SEM (MASEM), is a powerful tool for examining the relationship among latent constructs as well as those between the latent constructs and the manifest variables. Three classes of methods have been proposed for MASEM: (1) generalized least squares (GLS) in combining correlations and in estimating the structural model, (2) normal-distribution-based maximum likelihood (ML) in combining the correlations and then GLS in estimating the structural model (ML-GLS), and (3) ML in combining correlations and in estimating the structural model (ML). The current article shows that these three methods are equivalent. In particular, (a) the GLS method for combining correlation matrices in meta-analysis is asymptotically equivalent to ML, (b) the three methods (GLS, ML-GLS, ML) for MASEM with correlation matrices are asymptotically equivalent, (c) they also perform equally well empirically, and (d) the GLS method for SEM with the sample correlation matrix in a single study is asymptotically equivalent to ML, which has being discussed extensively in the SEM literature regarding whether the analysis of a correlation matrix yields consistent standard errors and asymptotically valid test statistics. The results and analysis suggest that a sample-size weighted GLS method is preferred for combining correlations and for MASEM.

The paper addresses the effect of external integration (EI) with transport suppliers on the efficiency of travel agencies in the tourism sector supply chains. The main aim is the comparison of different estimation methods used in the structural equation modeling (SEM), applied to discover possible relationships between EIs and efficiencies. The latter are calculated by the means of data envelopment analysis (DEA). While designing the structural equation model, the exploratory and confirmatory factor analyses are also used as preliminary statistical procedures. For the estimation of parameters of SEM model, three different methods are explained, analyzed and compared: maximum likelihood (ML) method, Bayesian Markov Chain Monte Carlo (BMCMC) method, and unweighted least squares (ULS) method. The study reveals that all estimation methods calculate comparable estimated parameters. The results also give an evidence of good model fit performance. Besides, the research confirms that the amplified external integration with transport providers leads to increased efficiency of travel agencies, which might be a very interesting finding for the operational management.

This study aimed to identify prognostic factors for severe sepsis-related in-hospital mortality using the structural equation model (SEM) analysis with statistical causality. Sepsis data from the Focused Outcomes Research in Emergency Care in Acute Respiratory Distress Syndrome, Sepsis, and Trauma study (FORECAST), a multicenter cohort study, was used. Forty seven observed variables from the database were used to construct 4 latent variables. SEM analysis was performed on these latent variables to analyze the statistical causality among these data. This study evaluated whether the variables had an effect on in-hospital mortality. Overall, 1148 patients were enrolled. The SEM analysis showed that the 72-hour physical condition was the strongest latent variable affecting mortality, followed by physical condition before treatment. Furthermore, the 72-hour physical condition and the physical condition before treatment strongly influenced the Sequential Organ Failure Assessment (SOFA) score with path coefficients of 0.954 and 0.845, respectively. The SOFA score was the strongest variable that affected mortality after the onset of severe sepsis. The score remains the most robust prognostic factor and can facilitate appropriate policy development on care.

Background: Telomere length (TL) maintenance plays an important role in bladder cancer (BC) and prognosis. However the manifold influence of everyday life exposures and genetic traits on leucocyte TL (LTL), is not fully elucidated. Methods: Within the framework of a hospital-based case (n = 96)/control (n = 94) study (all Caucasian males), we investigated the extent to which LTL and BC risk were modulated by genetic polymorphisms and environmental and occupational exposures. Data on lifetime smoking, alcohol and coffee drinking, dietary habits and occupational exposures, pointing to aromatic amines (AAs) and polycyclic aromatic hydrocarbons (PAHs) were collected. Structural equation modelling (SEM) analysis appraised this complex relationships. Results: The SEM analysis indicates negative direct links (p < 0.05) between LTL with age, DNA adducts, alcohol and NAT2, and positive ones with coffee, MPO and XRCC3; and between BC risk (p < 0.01) with cigarettes, cumulative exposure to AAs and coffee, while are negative with LTL and age. There was evidence of indirect effects (p < 0.05) on BC risk, probably via LTL reduction, by age and NAT2 (positive link), MPO and XRCC3 (negative link). Conclusions: Our study supports evidence that LTL attrition is a critical event in BC. The new finding that LTL erosion depends on some preventable everyday life exposures genetically modulated, opens new perspectives in BC prevention.

The Role of Providers in Mental Health Services Offered to American-Indian Youths

Socioeconomic status (SES) is inversely related to health status. Disparities in health status among races and ethnic groups are partly attributable to differences in SES, but the indirect pathways by which SES may influence health status are not widely studied. Using the Health and Retirement Study (HRS) data, this dissertation examined the pathways by which SES, via social, psychological, and behavioral factors predicted physical impairment and overnight hospitalization, and asked whether these indirect relationships differed by race/ethnicity. The HRS is a nationally representative multistage area probability sample administered biennially to respondents over the age of 51 and their spouses. Data collected between 2002 and 2010, covering five waves of the original HRS cohort born between the years 1931 and 1941, were used. Two analysis approaches, Structural Equation Modeling (SEM) and Reconstructability Analysis (RA), were used. Adjustments for the complex survey design were made in the SEM analysis, whereas, data were matched for the RA method using propensity scores. Results of the SEM analyses supported most of the hypothesized indirect relationships between SES variables and physical impairment via social and psychological factors, but the indirect effect of SES on physical impairment via behavioral factors was weak. Multiple group analyses of path equality using nested chi-square tests indicated that the indirect effect of SES on physical impairment status did not vary by race/ethnicity. Social, psychological and behavioral factors were weakly related to overnight hospitalization, and SES was not indirectly related to overnight hospitalization. While these results supported several hypothesized indirect relationships between SES variables and physical health status, the indirect effect sizes were small. However, because this study examined predictive paths across groups rather than compare mean differences, and because indirect effects are products of individual path coefficients, small effect sizes are not uncommon in mediation analysis. Moreover, over a lifetime, small effects may gradually add up increasing group differences in health status with greater benefits accruing to higher SES individuals via social and psychological factors, as observed in this study. The RA results showed that indirect relationships between SES and physical impairment were similar across races/ethnicities for identical variables with a few exceptions. In several cases, however, selected SES variables related to social and psychological variables were different for different groups. Cross-sectional indirect relationships were stronger than longitudinal indirect relationships. As in the SEM study, SES was not related to physical impairment via behavioral factors; and, across groups, SES was also

Mediation and conditional process analyses have become popular approaches for examining the mechanisms by which effects operate and the factors that influence them. To estimate mediation models, researchers often augment their structural equation modeling (SEM) analyses with additional regression analyses using the PROCESS macro. This duality is surprising considering that research has long acknowledged the limitations of regression analyses when estimating models with latent variables. In this article, we argue that much of the confusion regarding SEM’s efficacy for mediation analyses results from a singular focus on factor-based methods, and there is no need for a tandem use of SEM and PROCESS. Specifically, we highlight that composite-based SEM methods overcome the limitations of both regression and factor-based SEM analyses when estimating even highly complex mediation models. We further conclude that composite-based SEM methods such as partial least squares (PLS-SEM) are the preferred and superior approach when estimating mediation and conditional process models, and that the PROCESS approach is not needed when mediation is examined with PLS-SEM.

Even though data sets in psychology are seldom normal, the statistics used to evaluate covariance structure models are typically based on the assumption of multivariate normality. Consequently, many conclusions based on normal theory methods are suspect. In this paper, we develop test statistics that can be correctly applied to the normal theory maximum likelihood estimator. We propose three new asymptotically distribution-free (ADF) test statistics that technically must yield improved behaviour in samples of realistic size, and use Monte Carlo methods to study their actual finite sample behaviour. Results indicate that there exists an ADF test statistic that also performs quite well in finite sample situations. Our analysis shows that various forms of ADF test statistics are sensitive to model degrees of freedom rather than to model complexity. A new index is proposed for evaluating whether a rescaled statistic will be robust. Recommendations are given regarding the application of each test statistic.

본 연구는 공무원들의 스트레스나 피로수준과 같은 사회심리적 건강상태가 그들의 직급, 근무경력 등의 직업관련 특성이나 업무요구도, 업무자율성, 직장 내에서의 사회적 지지와 같은 직무스트레스 내용에 의해 영향을 받고 있는지를 구조방정식모형을 통하여 검토하고자 시도하였다. 연구대상은 대전광역시의 5개 정부 공공기관에 근무하는 일반 사무직 행정공무원 805명이었으며, 자료 수집은 2009년 10월 1일부터 11월 30일까지의 기간 동안에 표준화된 자기기입식 설문지(self-administered questionnaire)를 사용하여 실시하였다. 연구결과, 공무원들의 스트레스 및 피로수준은 직급, 근무경력, 업무만족도, 업무자율성, 상사의 지지도 및 동료의 지지도와는 유의한 음의 상관관계를 나타낸 반면, 년 간 결근일수, 업무요구도와는 유의한 양의 상관관계를 나타내었다. 구조방정식모형 분석 결과 연구대상자들의 직무스트레스 요인은 직업관련 특성 요인보다 사회심리적 건강상태에 더 큰 영향을 미쳤으며, 직무스트레스가 높을수록 사회심리적 건강상태를 나타내는 스트레스와 피로수준을 높여주는 것으로 나타났다. 한편, 직업관련 특성의 인지정도가 높을수록 직무스트레스 요인을 감소시키는 효과가 있는 것으로 나타났다. 따라서 공무원들의 스트레스나 피로를 감소시키기기 위해서는 직업관련 특성에 따른 직무내용의 개선을 위한 프로그램의 개발 및 시행이 필요할 것으로 생각된다. 【The effects of job related factors on psychosocial health status were examined by taking job stress into account. The self-administered questionnaires were performed, during the period between Oct. 1st, 2009 and Nov. 30th, 2009, to 805 individuals in the service of general administration located in Daejeon City. Causal relationships between job related factors, job stress, and, psychosocial health status were examined by structural equation model. The main finding as follows: For correlations of such indices of psychosocial health status as PWI and MFS with various studied items, all of these two indices were found to be in a negative correlation with the job position, job career, satisfaction in job life, autonomy of job, supervisors support and coworkers support, whereas in positive correlation with the experience of sick absence per year and job demand. With the analysis of structural equation model, the degree of job stress factors was more influential on the level of psychosocial health status than job related factors in such a manner as that the greater the job stress, while the lower the job related factors, the higher the psychosocial health status(PWI and MFS). Moreover, it was found to have the inter-relational effects that the greater the job related factors, the lower the job stress factors. Thus the effective strategy for stress and fatigue symptoms reduction among governmental employees requires additional programs focusing on innovated job specifications.】

Structural equation models (SEMs) can be estimated using a variety of methods. For complete normally distributed data, two asymptotically efficient estimation methods exist: maximum likelihood (ML) and generalized least squares (GLS). With incomplete normally distributed data, an extension of ML called “full information” ML (FIML), is often the estimation method of choice. An extension of GLS to incomplete normally distributed data has never been developed or studied. In this article we define the “full information” GLS estimator for incomplete normally distributed data (FIGLS). We also identify and study an important application of the new GLS approach. In many modeling contexts, the variables in the SEM are linear composites (e.g., sums or averages) of the raw items. For instance, SEMs often use parcels (sums of raw items) as indicators of latent factors. If data are missing at the item level, but the model is at the composite level, FIML is not possible. In this situation, FIGLS may be the only asymptotically efficient estimator available. Results of a simulation study comparing the new FIGLS estimator to the best available analytic alternative, two-stage ML, with item-level missing data are presented.

Comparative analysis of random forest, exploratory regression, and structural equation modeling for screening key environmental variables in evaluating rangeland above-ground biomass

A two-stage estimation procedure is developed to analyze structural equation models of polytomous variables based on incomplete data. At the first stage, the partition maximum likelihood approach is used to obtain the estimates of the elements in the correlation matrix. It will be shown that the asymptotic distribution of these estimates is jointly multivariate normal. The second stage estimates the structural parameters in the correlation matrix by the generalized least squared approach with a correctly specified weight matrix. Asymptotic properties of the second stage estimates are also provided. Extension of the theory to multisample models, and some illustrative examples are also included.