abstract: A simulation study was conducted to explore the influence of partial loading invariance and partial intercept invariance on the latent mean comparison of the second-order factor within a higher-order confirmatory factor analysis (CFA) model. Noninvariant loadings or intercepts were generated to be at one of the two levels or both levels for a second-order CFA model. The numbers and directions of differences in noninvariant loadings or intercepts were also manipulated, along with total sample size and effect size of the second-order factor mean difference. Data were analyzed using correct and incorrect specifications of noninvariant loadings and intercepts. Results summarized across the 5,000 replications in each condition included Type I error rates and powers for the chi-square difference test and the Wald test of the second-order factor mean difference, estimation bias and efficiency for this latent mean difference, and means of the standardized root mean square residual (SRMR) and the root mean square error of approximation (RMSEA). When the model was correctly specified, no obvious estimation bias was observed; when the model was misspecified by constraining noninvariant loadings or intercepts to be equal, the latent mean difference was overestimated if the direction of the difference in loadings or intercepts of was consistent with the direction of the latent mean difference, and vice versa. Increasing the number of noninvariant loadings or intercepts resulted in larger estimation bias if these noninvariant loadings or intercepts were constrained to be equal. Power to detect the latent mean difference was influenced by estimation bias and the estimated variance of the difference in the second-order factor mean, in addition to sample size and effect size. Constraining more parameters to be equal between groups—even when unequal in the population—led to a decrease in the variance of the estimated latent mean difference, which increased power somewhat. Finally, RMSEA was very sensitive for detecting misspecification due to improper equality constraints in all conditions in the current scenario, including the nonzero latent mean difference, but SRMR did not increase as expected when noninvariant parameters were constrained. / Dissertation/Thesis / Masters Thesis Educational Psychology 2016

abstract: In order to analyze data from an instrument administered at multiple time points it is a common practice to form composites of the items at each wave and to fit a longitudinal model to the composites. The advantage of using composites of items is that smaller sample sizes are required in contrast to second order models that include the measurement and the structural relationships among the variables. However, the use of composites assumes that longitudinal measurement invariance holds; that is, it is assumed that that the relationships among the items and the latent variables remain constant over time. Previous studies conducted on latent growth models (LGM) have shown that when longitudinal metric invariance is violated, the parameter estimates are biased and that mistaken conclusions about growth can be made. The purpose of the current study was to examine the impact of non-invariant loadings and non-invariant intercepts on two longitudinal models: the LGM and the autoregressive quasi-simplex model (AR quasi-simplex). A second purpose was to determine if there are conditions in which researchers can reach adequate conclusions about stability and growth even in the presence of violations of invariance. A Monte Carlo simulation study was conducted to achieve the purposes. The method consisted of generating items under a linear curve of factors model (COFM) or under the AR quasi-simplex. Composites of the items were formed at each time point and analyzed with a linear LGM or an AR quasi-simplex model. The results showed that AR quasi-simplex model yielded biased path coefficients only in the conditions with large violations of invariance. The fit of the AR quasi-simplex was not affected by violations of invariance. In general, the growth parameter estimates of the LGM were biased under violations of invariance. Further, in the presence of non-invariant loadings the rejection rates of the hypothesis of linear growth increased as the proportion of non-invariant items and as the magnitude of violations of invariance increased. A discussion of the results and limitations of the study are provided as well as general recommendations. / Dissertation/Thesis / Ph.D. Psychology 2013

Culture, as an advanced form of social life, is internalized within each individual as an essential component of learning, socializing, and developing (Baumeister, 2011; Greenfield et al., 2003). Self-regulated learning (SRL), as demonstrated in the literature, is essential for students’ academic success, where self-regulated learners strategically and metacognitively plan, monitor, and adapt their learning processes to achieve their goals in learning (Winne & Hadwin, 1998; Winne, 1995; Zimmerman, 2002). Because SRL theories significantly emphasize the importance of social contexts, culture is likely to influence how individuals develop and gain SRL competency. Nevertheless, there is a paucity of cross-cultural studies of SRL research; thus, this study aimed to examine and compare Chinese international and Canadian domestic students’ self-reported engagement in SRL processes and their academic performance. To achieve this purpose, we adopted an emic approach by evaluating Winne and Hadwin’s (1998) model of SRL and systematically comparing it with Chinese conceptualization of learning (e.g., Confucianism). Then, we used an advanced statistical method to investigate the measurement invariance of the Regulation of Learning Questionnaire (RLQ) designed to capture SRL as dynamic processes unfolding over time for Chinese and Canadian groups. Our findings supported configural and metric invariances across Chinese and Canadian cultural groups. Based on the evidence of partial scalar invariance, we also identified single items that contributed to scalar non-invariance. This study demonstrated the significance of examining the measurement invariance across cultures, which warrants comparability in cross-cultural comparisons, and contributed greatly to the current literature on the relation between culture and SRL. / Graduate

Studies have been conducted in both the United States (US) and Argentina to validate the Spanish version of the Gamblers’ Beliefs Questionnaire (GBQ-S). While similar factor structures of the GBQ-S were found in these studies, more detailed measurement invariance has not been explored. Determining measurement invariance across cultural groups would aid researchers in understanding what types of comparisons on latent constructs can be validly made between cultures using the GBQ-S. A secondary analysis was completed with data from two prior studies in Spanish-speaking samples from the US and Argentina. A modified baseline model that captured the two latent factors of Illusion of Control and Luck/Perseverance was selected for invariance testing. The evaluation of measurement invariance within a structural equation modeling framework established configural and metric, but not scalar, invariance. The GBQ-S can be validly used to measure the same latent variables across groups with comparable strengths between its items and factors; however, latent mean comparisons across groups may require further measurement refinement. The GBQ-S demonstrated notable degrees of measurement invariance between markedly different samples, providing further evidence for the existence of similar constructs and portability of the GBQ-S across populations.

The poor retention of special education teachers negatively impacts student academic outcomes. Special education teachers commonly cite burnout as a primary reason for leaving the field; however, there is a deficit of literature available to validate claims concerning special educators and their level of burnout. This study analyzed the psychometric properties of the Maslach Burnout Inventory: Educators' Survey using a sample of 349 special education teachers from schools across the nation (201 resource room special education teachers and 148 self-contained special education teachers). The Maslach Burnout Inventory: Educators' Survey measures three factors (e.g. subscales) emotional exhaustion, depersonalization, and personal accomplishment. A confirmatory factor analysis, an exploratory factor analysis, and a multi-group measurement invariance confirmatory factor analysis were conducted. The results showed measurement invariance between the two groups of teachers. During the exploratory factor analysis, a significant fourth factor, collaborative stress, emerged. These findings suggest the current factor structure of the Maslach Burnout Inventory: Educators' Inventory needs to be modified when measuring burnout among special education teachers. This includes the need to further explore how collaboration stress relates to special education teachers and how to implement formative collaboration practices to retain special education teachers.

Current recommended cutoffs for determining measurement invariance have typically derived from simulation studies that have focused on multigroup confirmatory factor analysis, often using continuous data. These cutoffs may be inappropriate for ordered categorical data in a longitudinal setting. This study conducts two Monte Carlo studies that evaluate the performance of four popular model fit indices used to determine measurement invariance. The comparative fit index (CFI), Tucker-Lewis Index (TLI), and root mean square error of approximation (RMSEA) were all found to be inconsistent across various simulation conditions as well as invariance tests, and thus were not recommended for use in longitudinal measurement invariance testing. The standardized root mean square residual (SRMR) was the most consistent and robust fit index across simulation conditions, and thus we recommended using ≥ 0.01 as a cutoff for determining longitudinal measurement invariance with ordered categorical indicators.

Magister Artium (Psychology) - MA(Psych) / While research into children’s subjective well-being (SWB) has advanced over the past decade, there is a paucity of cross-cultural research, particularly in South Africa. Moreover, while the adaptation and validation of instruments in English and Afrikaans are evident, other language groups have not received much attention. This study aimed to provide structural validation of the Students’ Life Satisfaction Scale across a sample of children in South Africa using multi-group analysis across three language groups (Setswana, Xitsonga, and Tshivenda). Within this process, the study aimed to use multi-group confirmatory factor analysis (MGCFA) to compare the structural validity and measurement invariance of the three language groups. Finally, the study aimed to determine the convergent validity of the three language groups of the SLSS by regressing them onto the single-item Overall Life Satisfaction Scale (OLS). The study uses data from Wave 3 of the South African Children’s Worlds Study and included a sample of 625 children across the language groups (Setswana: n = 187; Sesotho: n = 170; and Tshivenda: n = 268). For the overall pooled sample an excellent fit was obtained for a single-factor model, including one error-covariance. Standardised regression weights of the items ranged between .43 and .73. MGCFA revealed an acceptable fit for the configural model (unconstrained loadings); however, metric (constrained loadings) and scalar invariance (constrained loadings and intercepts) was not tenable. However, through the application of partial constraints metric invariance was tenable when Item 5 (I like my life) was freely estimated, while scalar invariance was tenable when Item 1 (I enjoy my life) and Item 5 (I like my life) were freely estimated. The results suggest that the Items: My life is going well; I have a good life; The things in my life are excellent; and I am happy with my life, are comparable by correlations, regression coefficients, and latent mean scores across the three language groups. Convergent validity using the OLS was obtained for the pooled sample and across the language groups. The key contribution of the study is establishing that the Setswana, Sesotho, and Tshivenda translated and adapted versions of the SLSS are valid for use within the South African context to measure children’s SWB, and that they can be grouped together in an overall pooled sample.

A restorative justice alternative to crime prevention termed reintegrative shaming theory by Braithwaite has seen increased attention as an alternative to retributive justice, although empirical investigations of its efficacy are limited. The purpose of the present study was to test confirmatory measurement and structural models of reintegrative shaming theory in order to assess the underlying theoretical model and the application of this theory in response to drunk driving offenses. Nine latent constructs were included in these models: reintegration, stigmatization, perceived fairness, self esteem, shame-guilt, embarrassment-exposure, unresolved shame, offender responsibility, and family support. Multigroup structural equation modeling was used to assess for measurement invariance of indicators used to measure these nine latent constructs between 724 drunk driving offenders randomly assigned to traditional court processing versus offenders assigned to reintegrative shaming conferencing following arrest. Partial metric and partial scalar invariance were found. Thus, analyses proceeded by conducting tests for significant differences in the latent means between groups. Offenders assigned to conferencing reported significantly higher mean values on the constructs reintegration, perceived fairness, self-esteem, shame-guilt, and family support, supporting Braithwaite's theory. Finally, a structural model was hypothesized based upon Braithwaite's theory to assess the relationships between the latent constructs. Three additional structural paths were included to achieve an acceptable model fit. This structural model was found to be partially invariant between groups. As predicted, a higher level of reintegration was associated with greater perceived fairness, while a higher level of stigmatization was related to decreased self-esteem and lower perceived fairness. In turn, greater self-esteem and perceived fairness were significantly related to higher reported experiences of shame-guilt and lower ratings of embarrassment exposure. Greater perceived fairness also corresponded to lower reported unresolved shame. Finally, greater shame-guilt was significantly related to greater offender responsibility and family support, while unresolved shame was significantly related to less offender responsibility acceptance. The findings from the current study support Braithwaite's hypotheses regarding the importance and benefits of disapproving of the criminal act and not the person, while allowing offenders to accept responsibility for their actions and attempt to remediate the wrong that they committed.

Indiana University-Purdue University Indianapolis (IUPUI) / Importance: Despite its widespread use in clinical settings and in behavioral medicine research, little is known about the psychometric performance of the PHQ-9 across major U.S. sociodemographic groups. Thus, utilizing a large sample representative of the U.S. population and confirmatory factor analysis (CFA), we determine the factor structure and measurement invariance of the PHQ-9 across groups based on sex, race/ethnicity, and education level. Objective: Our objective was to address key knowledge gaps by definitively determining the factor structure and measurement invariance of the PHQ-9 across major U.S. sociodemographic groups based on sex, race/ethnicity, and education level. Design: The continuous National Health and Nutrition Examination Survey (NHANES) is a cross-sectional, epidemiologic study designed to assess the health and nutritional status of the U.S. population. We examined data from the 2005-2014 survey years. Setting: NHANES is uses a stratified multistage probability sampling approach to enroll civilian, non-institutionalized adults and children in the U.S. Participants: For our final sample, we selected the 26,202 adult respondents with no missing PHQ-9 data. The factors of interest were sex (49.3% men, 50.7% women), race/ethnicity (48.9% non-Hispanic White, 23.7% non-Hispanic Black, 17.8% Mexican American, 9.7% other Hispanic), and education level (9.9% less than 9th grade, 16.6% 9th-12th grade but no diploma, vii 23.7% high school graduate/GED or equivalent, 28.9% some college or Associate’s degree, 20/8% college graduate or above). Main Outcome(s) and Measure(s): The Patient Health Quessionnaire-9 (PHQ-9) Results: Results revealed that the best solution for the PHQ-9 consists of a cognitive/affective factor (items 1. anhedonia, 2. depressed mood, 6. feelings of worthlessness, 7. concentration difficulties, 8. psychomotor disturbances, and 9. thoughts of death) and a somatic factor (items 3. sleep disturbance, 4. fatigue, and 5. appetite changes; RMSEA = 0.034, RMSEA 90% CI = 0.032–0.036, TLI = 0.984, CFI = 0.988). To evaluate measurement invariance, we then conducted single-group and multiple-group CFAs to carry out the 5 steps of measurement invariance testing. Dimensional, configural, weak factorial, strong factorial, and strict factorial invariance was established for the PHQ-9 across the sex, race/ethnicity, and education level groups, as all models demonstrated close fit and the ΔCFI was < 0.010 for all steps. Conclusions and Relevance: Using a U.S. representative sample, we determined that a two-factor solution for the PHQ-9 with a cognitive/affective factor and a somatic factor is invariant across sex, race/ethnicity, and education level groups. Therefore, clinically, the PHQ-9 is an acceptable measure to utilize in major U.S. sociodemographic groups, extending the use of this depression screener from the primary care clinic to the community. Additionally, we show that PHQ-9 cognitive/affective and somatic subscale scores have the same meaning and can be compared across major U.S. sociodemographic groups and provide a consistent, evidence-based approach to computing PHQ-9 subscale scores to be used in future studies.

No description available.