Impact of maternal alcoholism on separation of children from their mothers: findings from a sample of incarcerated women

Research on achievement goal orientation in sport has relied primarily on traditional statistical methodology to examine group mean differences. Unfortunately, examination of the measurement model is generally ignored prior to means testing. This study reports an application of structural equation modeling (SEM) in testing measurement invariance and latent mean structure of the Task and Ego Orientation in Sport Questionnaire (TEOSQ; Duda & Nicholls, 1989) using male and female college students. A confirmatory factor analysis for testing invariance revealed invariant measurement properties and factor structures across gender, indicating that task and ego orientation are similarly conceptualized by male and female students. Subsequent testing of latent mean structures, however, showed significant gender differences with respect to ego orientation, but no difference in task orientation. The SEM procedures used in the present study demonstrate additional construct validity and internal consistency reliability for the TEOSQ and, by confirming its factor structure, provide a sound psychometric basis for its continued use in substantive studies focusing on the comparison of achievement goal orientation across gender.     

The Role of Nonlinear Factor-to-Indicator Relationships in Tests of Measurement Equivalence.

Measurement invariance is a necessary condition for the evaluation of factor mean differences over groups or time. This article considers the potential problems that can arise for tests of measurement invariance when the true factor-to-indicator relationship is nonlinear (quadratic) and invariant but the linear factor model is nevertheless applied. The factor loadings and indicator intercepts of the linear model will diverge across groups as the factor mean difference increases. Power analyses show that even apparently small quadratic effects can result in rejection of measurement invariance at moderate sample sizes when the factor mean difference is medium to large. Recommendations include the identification of nonlinear relationships using diagnostic plots and consideration of newly developed methods for fitting nonlinear factor models. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Race, job applicants, and the Five-Factor Model of Personality: Implications for Black psychology, industrial/organizational psychology, and the Five-Factor Theory.

We tested the fit of the five-factor model of personality with a sample of African American (n?=?184) and Caucasian (n?=?168) job applicants using confirmatory factor analysis with tests of invariance across groups. Indicators for the analyses were responses to the 80 Bipolar Adjective Checklist. The results provided moderate support for the five-factor theory for both groups, and the addition of corresponding constraints on the factor loadings, factor correlations, and latent means did not lead to a significant loss in model fit. There were only differences on four elements of the error matrices. Thus, for the most part, the five-factor model fit equally well for African American and Caucasian applicants. However, for both groups, all factors were highly intercorrelated consistent with an "ideal responding" response set. Implications for Black psychology, I/O psychology, and the five-factor theory are discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Confirmatory factor analysis and item response theory: Two approaches for exploring measurement invariance.

Investigated the utility of confirmatory factor analysis (CFA) and item response theory (IRT) models for testing the comparability of psychological measurements. Both procedures were used to investigate whether mood ratings collected in Minnesota and China were comparable. Several issues were addressed. The 1st issue was that of establishing a common measurement scale across groups, which involves full or partial measurement invariance of trait indicators. It is shown that using CFA or IRT models, test items that function differentially as trait indicators across groups need not interfere with comparing examinees on the same trait dimension. Second, the issue of model fit was addressed. It is proposed that person-fit statistics be used to judge the practical fit of IRT models. Finally, topics for future research are suggested. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Structural invariance of cognitive abilities across the adult life span: A cross-sectional study.

This study examined the hypothesis that psychometric tests retain equivalent factor structures across samples widely differing in age. We estimated a best-fitting measurement model for 17 psychometric tests covering the 5 primary abilities of Inductive Reasoning, Spatial Orientation, Verbal Ability, Numerical Ability, and Perceptual Speed, using a sample of 1,621 participants (ages 22 to 95) from the 5th wave of the Seattle Longitudinal Study. We disaggregated the participants into 9 subsets (M ages?=?29, 39, 46, 53, 60, 67, 74, 81, and 90) and tesed the fit of the accepted model for each subset. We confirmed configural invariance for all subsets, but could not establish either complete or incomplete metric invariance for any set. These results confirm the stability of factor patterns across age but indicate serious limitations for valid cross-age comparisons of individual markers of psychometric abilities in age-comparative studies. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Longitudinal invariance of adult psychometric ability factor structures across 7 years.

The hypothesis that psychometric ability tests retain equivalent factor structures across a 7-year interval was examined in a sample of 984 persons (disaggregated into 6 cohort groups: M ages at first test?=?32, 46, 53, 60, 67, and 76), assessed in 1984 and 1991 as part of the Seattle Longitudinal Study. A best fitting measurement model was estimated for 20 psychometric tests marking the 6 primary abilities of Inductive Reasoning, Spatial Orientation, Perceptual Speed, Numeric Facility, Verbal Ability, and Verbal Recall. Gender was partialed out at the variable level by including a gender factor. Weak factorial invariance over time was demonstrated for all cohorts. Configural invariance could be demonstrated across all cohort groups. However, weak factorial invariance across groups could be accepted for all but the youngest and oldest groups. Latent means were modeled for the accepted solutions across time and cohort groups. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Confirmatory factor analysis of the styles of handling interpersonal conflict: First-order factor model and its invariance across groups.

Confirmatory factor analysis of data (from 5 samples, n?=?484 full-time employed management students; n?=?550 public administrators; n?=?214 university administrators; n?=?250 bank managers and employees in Bangladesh; and n?=?578 managers and employees) on the 28 items of the Rahim Organizational Conflict Inventory—II were performed with LISREL 7. The results provided support for the convergent and discriminant validities of the subscales measuring the 5 styles of handling interpersonal conflict (integrating, obliging, dominating, avoiding, and compromising) and general support for the invariance of the 5-factor model across referent roles (i.e., superiors, subordinates, and peers), organizational levels (top, middle, lower, and nonmanagement), and 4 of the 5 samples. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Evaluating the Impact of Partial Factorial Invariance on Selection in Two Populations.

Studies of factorial invariance examine whether a common factor model holds across multiple populations with identical parameter values. Partial factorial invariance exists when some, but not all, parameters are invariant. The literature on factorial invariance is unclear about what should be done if partial invariance is found. One approach to this problem evaluates the impact of partial invariance on accuracy of selection on the basis of a composite of the measures whose factor structure is being studied. Assuming a single-factor model holds, accuracy of selection using the composite is evaluated under varying degrees of partial invariance. A variety of examples are presented with discussion of extensions and limitations. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Invariance and convergent and discriminant validity between mothers' and fathers' ratings of oppositional defiant disorder toward adults, ADHD-HI, ADHD-IN, and academic competence factors within Brazilian, Thai, and American children.

Confirmatory factor analysis was used to test the invariance of an oppositional defiant disorder toward adults, attention-deficit/hyperactivity disorder-hyperactivity/impulsivity, attention-deficit/hyperactivity disorder-inattention, and an Academic Competence factor model between mothers' and fathers' ratings within Brazilian (n = 894), Thai (n = 2,075), and American (n = 817) children with the Child and Adolescent Disruptive Behavior Inventory (G. L. Burns, T. Taylor, & J. Rusby, 2001a, 2001b). The results showed invariance of item loadings, intercepts, and residuals, as well as factor variances, covariances, and means between mothers' and fathers' ratings within each sample. Convergent and discriminant validity was also observed for the between-parent factor correlations, thus providing additional support for the construct validity of the Child and Adolescent Disruptive Behavior Inventory. The confirmatory factor analysis invariance procedure provides a much better way to examine between-source ratings of behavior problems in children than do the simple correlation and raw discrepancy score procedures. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Relationship of Children's Depression Inventory factors to major depression among adolescents.

This study examined the relationship between the Children"s Depression Inventory (CDI) scores and major depression, conduct disorder, and anxiety disorder diagnoses. Participants were 107 (58 male, 49 female) psychiatric inpatients, aged 12–18 years (M?=?15.4, SD?=?1.5). Definite major depression participants (n?=?26) reported higher scores than nondepressed participants (absence of any depression diagnosis, n?=?81) on all 5 CDI factor scores and the total CDI score. Conduct disorder participants scored higher on the externalizing factor; no other significant main or interaction effects were obtained for conduct disorder or anxiety. A discriminant function model using the 5 CDI factor scores classified participants as depressed versus nondepressed with a high degree of accuracy; a model using only the CDI total score yielded comparable discriminatory accuracy. The CDI total score was recommended as the most practical measure for classifying participants as depressed or not depressed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Methodological Challenges in Union Commitment Studies.

Methodological problems in studies of union commitment were identified and illustrated with data from 4,641 members and 479 stewards in 297 local teachers' unions. Using a 20-item union commitment scale, results confirmed the existence of 3 substantive factors and 1 method factor at the individual level of analysis: loyalty to the union, responsibility to the union, willingness to work for the union, and a factor of negatively worded items. Tests of measurement invariance showed that the scale captured commitment for rank-and-file members but not for union stewards. The authors also found partial measurement invariance between long-time and newer members and full measurement invariance between men and women. Finally, the authors found that violation of the statistical assumption of independence reduced model fit when individual commitment scores were analyzed without attention to the hierarchical nature of the data. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Measuring subjective life satisfaction in persons with severe and persistent mental illness: A measurement quality and structural model analysis.

This study investigated measurement of subjective quality of life by Lehman's Quality of Life Interview (QOLI). Participants from 2 states were included (N?=?502). Confirmatory factor analysis conducted on 29 QOLI subjective satisfaction items contrasted first- and second-order factor models of domain structure in fit of observed data. Both models yielded adequate fits, but the second-order factor model provided a parsimonious explanation of first-order QOLI domain correlations. Factor structure and item loadings, estimated in Cincinnati, Ohio (N?=?285), were invariant in Baltimore, Maryland (N?=?217). Model testing on 1-year follow-up responses (N?=?372) indicated invariant parameter estimates over the follow-up period. Stability estimates were high, ranging from r?=?.20 for financial satisfaction to r?=?.73 for the second-order factor of general well-being. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

A multicultural assessment of adolescent connectedness: Testing measurement invariance across gender and ethnicity.

Counselors, psychologists, and evaluators of intervention programs for youth increasingly view the promotion of connectedness as an important intervention outcome. When evaluating these programs, researchers frequently test whether the treatment effects differ across gender and ethnic or racial groups. Doing so necessitates the availability of culturally and gender-invariant measures. We used the Hemingway: Measure of Adolescent Connectedness to estimate the factor structure invariance and equality of means across gender and 3 racial/ethnic groups with a large sample of middle school adolescents. From a practical perspective, the 10-scale model suggested factor structure invariance across gender and racial or ethnic (i.e., African American, Caucasian, and Latina/o) groups of adolescents. However, tests for partial invariance revealed some group difference on the factor loadings and intercepts between gender and ethnic/racial groups. When testing for mean equivalence, girls reported higher connectedness to friends, siblings, school, peers, teachers, and reading but lower connectedness to their neighborhoods. Caucasians reported higher connectedness to their neighborhoods and friends but lower connectedness to siblings than African Americans and Latinos. African Americans reported the highest connectedness to self (present and future) but lowest connectedness to teachers. Latinos reported the lowest connectedness to reading, self-in-the-present, and self-in-the-future. Overall, this study reveals racial/ethnic and gender mean differences on several connectedness subscales and suggests the Hemingway subscales are, from a practical perspective, invariant across gender and ethnicity and therefore appropriate for most assessment and evaluation purposes. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Structural invariance of vocational interests across Japanese and American cultures.

Two samples of college students (Japanese, N?=?373; American, N?=?401) were given the Inventory of Occupational Preferences (T. J. G. Tracey & J. Rounds, 1996a), and the structural invariance across cultures was examined at the item and at various scale levels. Correlations of principal-components factor loadings demonstrated that the item structure was similar across cultures. Three separate scale structures were examined: Holland's six-type circular model, Tracey and Round's eight-type circular model, and Tracey and Round's spherical representation. The fit of the six-type and spherical models to the Japanese sample was significantly worse than their fit to the U.S. sample. There were no differences in cross-cultural fit for the eight-type model. Few structure differences were found between and within genders. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Measurement invariance of the Illness Intrusiveness Ratings Scale's three-factor structure in men and women with cancer.

Objective: Illness- and treatment-related disruptions to valued activities and interests (illness intrusiveness) are central to quality of life in chronic disease and are captured by three subscales of the Illness Intrusiveness Ratings Scale (IIRS): the Instrumental, Intimacy, and Relationships and Personal Development subscales. Using individual (CFA) and multisample confirmatory factor analyses (MSCFA), we evaluated measurement invariance of the IIRS's 3-factor structure in men and women with cancer. Method: Men (n = 210) and women (n = 206) with 1 of 4 cancer diagnoses (gastrointestinal, head and neck, lymphoma, lung) recruited from outpatient clinics completed the IIRS. In the MSCFA, we applied an analysis of means and covariance structures approach to test increasingly stringent equality constraints on factor structure parameters to evaluate weak, strong, and strict measurement invariance of the 3-factor structure between men and women. Results: Individual CFAs demonstrated fit of the hypothesized 3-factor structure for men and women, although more consistently for men. The 3-factor structure was superior to an alternative 1-factor structure. MSCFA results indicated that parameters of the 3-factor structure could be considered equivalent between the sexes up to the level of strong invariance. Strict invariance was not supported. Conclusions: Overall, IIRS scores can be interpreted similarly for men and women with cancer. Illness intrusiveness can be considered as important in the psychosocial adaptation of people with cancer as it is for people affected by other chronic conditions. (PsycINFO Database Record (c) 2011 APA, all rights reserved)     

Measurement invariance in longitudinal clinical research assuming change from development and intervention.

Valid and reliable measures of psychological and behavioral constructs are critical to clinical research, particularly longitudinal treatment research, in which multiple groups are compared over time for possible changes within and between constructs as a result of intervention or development. Structural equation modeling (SEM) analysis is a class of statistical procedures that can be used to test multiple hypotheses about these relationships simultaneously while controlling for measurement error. The procedures have been applied primarily to testing between-construct relationships in nonexperimental studies, with relatively little emphasis on establishing whether measures are sufficiently invariant across groups and across time to permit these tests. This article uses an empirical example of a longitudinal experimental prevention study with 2 groups to illustrate the use of SEM, first, to systematically test measurement invariance across groups at each wave of measurement, and second, after establishing measurement invariance, to test structural invariance longitudinally. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Is aggression the same for boys and girls? Assessing measurement invariance with confirmatory factor analysis and item response theory.

Gender differences in aggression have typically been based on studies utilizing a mean difference method. From a measurement perspective, this method is inherently problematic unless an aggression measure possesses comparable validity across gender. Stated differently, establishing measurement invariance on the measure of aggression is prerequisite to making any inferences about gender differences. This paper investigated whether aggression functions in the same way across gender using teacher-rated aggression scales from the Behavior Assessment System for Children (BASC). The samples for this study consisted of the BASC standardization samples, including 1,223 children (628 boys, 595 girls) and 788 adolescents (379 boys, 409 girls). The measurement invariance of the BASC aggression scales was examined with both confirmatory factor analysis (CFA) and item response theory (IRT). CFA results showed that there was not enough evidence to support the measurement invariance of the scales across gender at a scale level; IRT results found that only a few items were significantly different across gender. The two methods have similarities and differences in studying measurement invariance, so they can be complementary to each other. The implications of the findings for research and practice in school psychology are discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

The Older Adult Disability Scale: Development and validation.

Examined the development and validity of the Older Adult Disability Scale (OADS), a self-report instrument designed to measure attitudes toward older adults (aged 60 yrs and older) with physical disabilities. Data from a community sample (n ?=?207) were used for alpha coefficients, subscale intercorrelations, demographic correlations, and factor analyses. Data from a reliability sample (n ?=?27) were used for a 2-week test-retest reliability check; and data from an identified "positive attitude" sample of health care professionals and students (n ?=?57) were used to test construct validity. Exploratory and confirmatory factor analyses were performed resulting in a 36-item instrument with four subscales of Cranky Dependency, Discomfort with Aging, Positive Attributes of Aging, and Negative Expectations of Disability. Satisfactory internal consistency, high test-retest reliability, and moderately high subscale intercorrelations were found. Validity was supported by the highly significant differences in OADS scores between the "positive attitude" and community samples. In conclusion, the OADS provides a promising measure of attitudes toward older adults with disabilities. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Invariance in measurement and prediction: Their relationship in the single-factor case.

Two forms of bias in psychological measurement are often distinguished. Predictive bias concerns systematic inaccuracies in predictions based on a test. Measurement bias concerns systematic inaccuracies in the information provided by a test about the latent variable or variables to be measured. Although both forms of bias are important, their relationship has not been clearly described in the literature. Their relationship is here described under a single-factor model for the predictor battery and the criterion. Several theorems establish the general inconsistency between factorial invariance and slope invariance (i.e., invariance of regression slopes). Predictor-criterion sets that are factorially invariant are not slope invariant under realistic conditions. The implications of this inconsistency for research on Spearman's "g" in employment testing, and for the general study of test bias, are discussed. (PsycINFO Database Record (c) 2011 APA, all rights reserved)