## No full-text available

To read the full-text of this research,

you can request a copy directly from the authors.

Advances in structural equation modeling techniques have made it possible to test models in data that are not normally distributed. This can lead to more realistic model testing in developmental psychology. Several alternate techniques are illustrated in structural equation models here in order to compare the results that are obtained. Maximum-likelihood and generalized least-squares estimators for normally distributed data are compared with Browne's asymptotically distribution-free technique for continuous nonnormally distributed data and Muthen's estimator for dichotomous indicators. While different critical ratios are found for some parameters estimated in models, the results are generally comparable so long as one does not consider absolute fit to be a critical factor in "accepting" a causal model as a good one.

To read the full-text of this research,

you can request a copy directly from the authors.

... Available evidence from studies using both simulated and empirical data (e.g., Boomsma, 1983;Browne, 1984;Huba & Harlow, 1987;Muthen & Kaplan, 1985;Tanaka, 1984;Tanaka & Huba, 1987;Windle, Barnes, & Welte, 1989) suggests that the accuracy of tests of statistical significance is adversely affected when normal-theory methods such as ML are inappropriately employed with nonnormal data. Although ML parameter esti-mates generally remain trustworthy even when the data are markedly nonnormal, the associated standard errors are incorrect, and the overall model chi-square may be inflated. ...

... The largest discrepancies were observed for those parameters reflecting the effects of constructs whose indicators were particularly poorly distributed. Although these findings are largely in agreement with the results of other studies that have examined this issue, the stability of the parameter estimates across the two methods was somewhat less than we expected on the basis of prior research (e.g., Huba & Harlow, 1987). It is informative to compare these differences with the differences that were observed across estimators for the model of the effects of confidant support. ...

Using a latent-variable modeling approach, relationships between social ties and depression were studied in a sample of 201 older adults. Both positive and negative ties were related to concurrent depression, whereas only negative ties predicted future depression. Nonnormally distributed scores were observed for several variables, and results based on maximum likelihood (ML), which assumes multivariate normality, were compared with those obtained using Browne's (1982, 1984) arbitrary distribution function (ADF) estimator for nonnormal variables. Inappropriate use of ML with nonnormal data yielded model chi-square values that were too large and standard errors that were too small. ML also failed to detect the over-time effect of negative ties on depression. The results suggest that the negative functions of social networks may causally influence depression and illustrate the need to test distributional assumptions when estimating latent-variable models.

... Figueredo, personal communication). there is literature support that ML estimation is robust under moderate violation of normality assumptions (Bentler, 1987(Bentler, , 1989Huba & Harlow, 1987). ...

... Additional instruments indexing self-transcendence also would have enhanced the measurement of that concept in the type of analysis used in this study. An advantage of factor analytic structural equations modeling is that measuring a concept with several indicators permits the use of the common meaning (covariance) of the indicators thereby eliminating the potentially confounding influence of measurement error found in the use of separate indicators (Huba & Harlow, 1987 (1-7) with word choices (e.g., "Not At All" and "Very Much") as was done with most of the other instruments in this study. The findings of this study apply only to women. ...

... We used maximum likelihood estimation because of its capacity to accommodate large models and its robustness to deviations from multivariate normality (Huba & Harlow, 1987). We evaluated goodness of fit using three measures of practical fit: delta, rho, and the comparative fit index (Bentler, 1990a). ...

This article provides an application of structural equation modeling to the evaluation of cross-lagged panel models. Self-reports of physical and mental health at 3 different time points spanning a 4-year interval were analyzed to illustrate the cross-lagged analysis methodology. Data were collected from a sample of 856 patients with hypertension, diabetes, heart disease, or depression (or any combination of these) participating in the Medical Outcomes Study. Cross-lagged analyses of physical and mental health constructs revealed substantial stability effects across time. A structural model with standard effects revealed positive effects of physical health on mental health but negative (suppression) effects of mental health on physical health. The effects of mental health on physical health became nonsignificant when the model was revised by adding nonstandard effects (direct effects of measured variable residuals on latent variables). Recommendations for structural equation modeling of cross-lagged panel data are provided.

... This assumption can lead to erroneous conclusions, and suspect inferences. Many researcher have considered SEM when normality is misspecified; see, e.g., Yuan and Bentler (1998); Huba and Harlow (1987); Gao et al. (2008). ...

In this paper, we propose and present a nonparametric data smoothing method via the kernel smoothing functions to make structural equation modeling (SEM) robust to a specific type of model misspecification, that is an incorrect distributional assumption. Although most statistical techniques are based on an implicit assumption of normality, real data often exhibits nonnormal kurtosis (heavily peaked), skewness, or both. These characteristics, if ignored, can make model identification difficult and inference not reliable. It is important to note that these are characteristics present in most real multivariate high-dimensional datasets. There is much recent study devoted to this type of misspecification. Using a large scale Monte Carlo simulation study, we evaluate the efficacy of our proposed approach in improving the frequency with which a correctly specified model is selected by information complexity criteria when the normality is misspecified. We also show our results on a benchmark reference real dataset to study the quality of life. Our results indicate that the data smoothing kernel transformation (KDS-SEM) leads to a better fitting structural equation model (SEM) and model selection.

... Firstly, normality tests were performed, and data were shown not to be normally distributed for any of the variables involved in the model. As a consequence of this, the UL-Unweighted Least Squares method was used through the LISREL 8.80 program, being this an option that does work quite well without the assumption of normality and under analytical conditions such as small samples or excessive kurtosis (Huba & Harlow, 1987). Obtained standardized results are provided ( Figure 2). ...

This paper shows the results of a study based on path analysis about the predictive and mediator role of environmental awareness on the conceptions of Pre-service Primary teachers about science education. The model thus obtained is validated through a representative sample of 300 students from a four-year university degree for teacher training. The method of unweighted least squares is used to perform the analysis and the results show a good fit to the data model. Therefore, it is assumed that students with higher levels of environmental awareness may be more effective in their future teaching profession. Consequently, the increase in environmental practices in schools makes a small but important contribution to the overall improvement of the environment and the training of future citizens.

... Two models (Figure 1) were tested using path analysis. Maximum likelihood method from LISREL 8.54 was employed because this procedure has been shown to be robust to departures from normality [37]. ...

Objective: Self-esteem has been identified as a strong predictor of depression and maladaptive behaviours in adolescents. Two relational models (A and B) were tested on the antecedent and consequent variables of self-esteem.
Methods: A representative sample of 610 Spanish adolescents (52% boys) ranging in age from 11 to 16 years of age (mean age = 13.38 years, SD = 1.70 years) was used. The participants completed a battery of instruments measuring self-esteem, self-concept, importance of self-concept domains, depression symptoms, and problem behaviour. Covariance structure analysis (path analysis) from LISREL 8.54 was employed.
Results: Model B presented a better fit (χ²difference (A–B) = 160.29, p < 0.001). Perceived social support (parents and classmates), and perceptions of competence in domains deemed important were positively related to self-esteem. While both self-esteem and social support (parents and classmates) were negatively related to depression symptoms, only social support (parents and teacher) appeared negatively related to behavioural problems.
Conclusions: While self-esteem constitutes a strong correlate of affective symptoms in Spanish adolescents, it is unrelated to maladaptive behaviour. Social support emerged as an important protective factor. Concrete proposals for future research and prevention are discussed.

... Thus, maximum likelihood estimation was considered appropriate to analyze our datasets, since our structural equations models are theory-driven and sample size is reasonably Vamvaka et al. Journal of Innovation and Entrepreneurship (2020) large (Curran, West, & Finch, 1996;DiLalla, 2000;Huba & Harlow, 1987;Olsson, Foss, Troye, & Howell, 2000;Wang, Fan, & Willson, 1996;Weston & Gore, 2006). ...

Abstract Drawing on the theory of planned behavior, the main purpose of this cross-sectional study was to identify gender-related differences in the levels of and the interrelations among attitude toward entrepreneurship, perceived behavioral control, and entrepreneurial intention using multi-group structural equations modeling in which the dimensions of these constructs were disentangled and treated as latent variables that were indirectly inferred from multiple indicators. The sample of the study consisted of 441 Greek tertiary education undergraduate information technology students. The results showed that attitude consists of two components—one instrumental and one affective; perceived behavioral control is comprised of two factors—perceived self-efficacy and perceived controllability; and entrepreneurial intention is best represented by three factors—choice intention, commitment to entrepreneurship, and nascent entrepreneurship. The findings indicated further that affective attitude and perceived self-efficacy are by far the strongest predictors of intention, thus highlighting the role of emotions in the entrepreneurial process. Our work revealed also that the relationship between commitment to entrepreneurship and nascent entrepreneurship is stronger in men than in women. Conceiving nascent entrepreneurship as a proxy for entrepreneurial behavior, this finding implies that gender is a moderator of the entrepreneurial intention-action translation. Despite its limitations, this study makes some important contributions and implications to the literature of entrepreneurship. These and future research suggestions are also discussed.

... Opšte poznata činjenica je da se na osnovu korelacionih podataka ne može zaključivati o kauzalnim relacijama. SE model (Connell & Tanaka, 1987;Biddle & Marlin, 1987;Mulaik, 1987;Tanaka, 1987;Huba & Harlow, 1987;Connell, 1987;Jöreskog & Sörbom, 1989) predstavlja tehniku koja, prevazilazeći ograničenja postojećih multivarijantnih tehnika, omogućava testiranje valjanosti nekog teorijskog modela kojim se definišu kauzalne i korelacione veze. U najopštijem tipu modela mogu se razlikovati indikatorske varijable (varijable koje se registruju u istraživanju) i hipotetički konstrukti (varijable koje su definisane sa više indikatirskih varijabli). ...

English version
STRUCTURE OF PARENTS' IMPLICIT KNOWLEDGE ABOUT PERSONAL CHARACTERISTICS OF THEIR CHILDREN
The basic aim of this work is to establish basic dimensions around which parents’ conception about personality of their children are organized. Identification of the basic dimensions enables researchers to describe and study parents’ implicit knowledge about personality of their children, as well as comparison between different parents, coherence of implicit knowledge of two parents, parent implicit representation of the ideal personality of the child at given age etc.
The first part of the thesis defines conceptual and methodological framework of the research. In this context, key conceptual issues related to studies of parents’ implicit representations of personality of their children have been discussed as well as their implicit understanding of child development and socialization. As a result of the discussion a classification of parents’ implicit representations has been suggested: (a) parents’ intuition about of development, (b) parents’ intuitions about socialization, and (c) parents’ representation of personality of their children. Finally, an analysis of contemporary researches in this field has suggested that parents’ representations of personality of their children is a topic that has been not very much studied previously.
In the part that follows, a nature of parents’ representations about the personality of their children has been analyzed and discussed. The analysis has shown that it is consisted of practical, implicit and intuitive knowledge which is structured and associated with relevant context. Relying on the results of this analysis, a complex, multi-phase method has been developed, which is enabling holistic description of the structure of parents’ representations about the characteristics of their children.
In the next section of the thesis the design of the research is presented and discussed. In the first phase of the research design, the analysis of typical terms and phrases used by parents to describe personality of their children is presented. The second phase is consisted of an analysis of meanings that parents relate to these terms and phrases. The third phase of the model describes selection of representative sample of terms and phrases which further has been used in the research. The fourth phase defines basic dimensions of intuitive parents’ representation about personality of their children. Five basic dimensions have been extracted: (1) responsibility, (2) aggressiveness, (3) dependency, (4) prosociality, and (5) narcism. In addition, it has been established that five basic dimensions are the same regardless from parents’ gender, as well as from gender and age of their children.
Serbian version
STRUKTURA IMPLICITNOG ZNANJA RODITELJA O OSOBINAMA SOPSTVENE DECE
Osnovni cilj ovog rada je da se utvrde bazične dimenzije oko kojih se organizuje intuitivna predstava roditelja o osobinama sopstvene dece. Ove dimenzije omogućavaju da se opiše implicitno znanje roditelja o sopstvenoj deci, da se upoređuju različiti roditelji, da se proverava tačnost roditeljske predstave itd.
U prvom, teorijsko-metodološkom delu rada definisan je konceptualno-metodološki okvir istraživanja. U tom kontekstu razmotrena je priroda roditeljskih predstava (koncepcija, verovanja, znanja) o deci, njihovom razvoju i socijalizaciji, njihovo poreklo i funkcija. Predložena je i klasifikacija roditeljskih verovanja: (a) laička shvatanja o razvoju, (b) laička shvatanja o socijalizaciji, i (c) predstava roditelja o osobinama sopstvene dece. Analiza dosadašnjih istraživanja pokazala je da su roditeljske predstave o osobinama sopstvene dece relativno neistražen segment roditeljskih verovanja.
Nakon toga, razmotrena je priroda roditeljskih predstava o ličnosti sopstvenog deteta. Utvrđeno je da se radi o praktičnom, implicitnom i intuitivnom znanju koje je uređeno i kontekstualizovano. Polazeći od toga, definisan je kompleksan višefazni metod koji omogućava da se stekne celovita predstava o prirodi i strukturi roditeljske predstave o osobinama sopstvenog deteta.
U istraživačkom delu rada prikazani su rezultati svake faze istraživanja. U prvoj fazi su analizirani termini i fraze kojima roditelji opisuju sopstvenu decu, a u drugoj fazi su analizirana značenja koja roditelji mogu pripisivati datim terminima i frazama. U trećoj fazi je izvršena selekcija reprezentativnog uzorka termina i fraza koji su bili korišćeni u daljim ispitivanjima. U četvrtoj fazi su izdvojene bazične dimenzije koje leže u osnovi intuitivne predstave roditelja o tome kakva su njihova deca. Izdvojeno je pet bazičnih dimenzija: (1) odgovornost, (2) agresivnost, (3) zavisnost, (4) prosocijalnost i (5) narcisoidnost. Pored toga, utvrđeno je da su ove dimenzije u osnovi roditeljskih predstava bez obzira da li je to predstava majke ili oca, da li se odnosi na dečake ili devojčice, i bez obzira na to koliko godina dete ima.

... However, the concerns may be mitigated somewhat by the use of a latent variable approach in our analysis. This allowed us explicitly to model measurement error and, as a consequence, the constructs and associated parameter estimates can be considered relatively error-free (Huba & Harlow, 1987). ...

Although regular physical activity is recommended for pregnant women, compared to pre-pregnancy, antenatal physical activity often reduces or ceases completely. Drawing from the theory of planned behavior, self-determination theory, and theory on self-control, we aimed to test an integrative model of physical activity in a sample of pregnant women. The current study was conducted in Brisbane, Australia, in 2014–2015 using a prospective-correlational design with a one-week follow-up. Participants (N = 207, Time 1; Meanage = 30.03 years, SDage = 4.49 years) completed an initial survey measuring: intrinsic motivation from the self-determination theory, social cognitive constructs from the theory of planned behavior, and self-control from the self-control theory, followed by a self-report measure of physical activity one-week later (n = 117, Time 2). A well-fitting structural equation model accounted for 73 and 42 percent of the variance in intention and physical activity behavior, respectively. Perceived behavioral control and attitude, but not subjective norm, mediated the effect of intrinsic motivation on intention. Intention, perceived behavioral control, and self-control were positively associated with physical activity behavior. Future behavioral interventions aiming to promote physical activity during pregnancy, a period when the physical activity levels typically decline, should consider the multiple processes advocated in the integrative model as necessary for motivated action.

... Maximum likelihood estimation was used to generate the standardized parameter estimates because it is robust in dealing with data that deviate from multivariate normality (Huba & Harlow, 1987). To determine the fit of the models to the observed data, we used the chi-square statistic (Bollen, 1989), the nonnormed fit index (NNFI; Tucker & Lewis, 1973), and the root-meansquare error of approximation (RMSEA) (Steiger, 1990). ...

Self-determination theory was applied to explore the motivational basis of adherence to long-term medication prescriptions. Adult outpatients with various diagnoses who had been on a medication for at least 1 month and expected to continue (a) completed questionnaires that assessed their autonomous regulation, other motivation variables, and perceptions of their physicians' support of their autonomy by hearing their concerns and offering choice; (b) provided subjective ratings of their adherence and a 2-day retrospective pill count during an interview with a clinical psychologist; and (c) provided a 14-day prospective pill count during a subsequent, brief telephone survey. LISREL analyses supported the self-determination model for adherence by confirming that patients' autonomous motivation for adherence did mediate the relation between patients' perceptions of their physicians' autonomy support and their own medication adherence.

... To provide a metric for the latent variables (LVs), one of the indicators loading for each variable was set to 1.00. Maximum likelihood estimation was used to generate standardized parameter estimates because it is robust in dealing with deviations from multivariate normality (Huba & Harlow, 1987). ...

American and Russian students' psychic health, extrinsic and intrinsic aspirations and parent and teacher perceptions are compared. It is found that the higher intrinsic aspirations the better psychic health characteristics are not, on the contrary, orientation on extrinsic ones leads to deterioration of psychic well-being. As to parent and teacher perceptions the latter influence students' life values and the former do not.

... There was little evidence of any substantial departures from nor-mality and most of the distributional characteristics for the individual items and scales were within acceptable limits. Any departures from normality were not sufficiently large enough to strain the robustness of the ML estimation procedure used in the SEM (Huba & Harlow, 1987). A careful inspection of the bivariate relations indicated a small lack of pretest equivalence. ...

This book, first published in 2002, represents a systematic discussion of the Gateway Hypothesis, a developmental hypothesis formulated to model how adolescents initiate and progress in the use of various drugs. In the United States, this progression proceeds from the use of tobacco or alcohol to the use of marijuana and other illicit drugs. This volume presents a critical overview of what is currently known about the Gateway Hypothesis. The authors of the chapters explore the hypothesis from various perspectives ranging from developmental social psychology to prevention and intervention science, animal models, neurobiology and analytical methodology. This volume is original and unique in its purview, covering a broad view of the Gateway Hypothesis. The juxtaposition of epidemiological, intervention, animal and neurobiological studies represents a new stage in the evolution of drug research, in which epidemiology and biology inform one another in the understanding of drug abuse.

... Because of the small sample, we used EQS's elliptical estimation to assess the models (Bentler & Chou, 1987). This method does not require normal variable kurtosis (Bentler, 1985), the absence of which poses a graver problem than skewed data (Huba & Harlow, 1987). Moreover, we report unstandardized parameter estimates (L. ...

Hom, Griffeth, and Sellaro's (1984) theoretical alternative to Mobley's (1977) turnover model was investigated in two studies. In Study 1, conceptual distinctions among model constructs and operationalizations of those constructs were validated. 206 nurses were surveyed, and constructs were assessed with multiple indicators. Although discriminating most constructs, structural equation modeling (SEM) identified a more parsimonious conceptualization in which a general construct underlies withdrawal cognitions. Other SEM analyses supported the indicators' construct validity and Hom et al.'s structural network. In Study 2, a longitudinal analogue of Hom et al.'s model was tested. A survey of 129 new nurses measured model constructs on three occasions. SEM disclosed that some causal effects in this model materialized contemporaneously, whereas others emerged after a lengthy time. Moreover, these causal effects systematically changed during newcomer assimilation. Implications for future research of turnover models are discussed.

... We used a structural equation modeling approach in Mplus 7.1 to examine associations between pubertal timing and tempo, spare time activities, and adolescent delinquency. This latent variable approach is superior to combining single items into (quasi-) manifest variables (e.g., using averaged scores) as it corrects for residual variance inherent in many constructs (resulting in unsatisfactory internal consistency) and allows for estimation of relationships among constructs that deviate from normality (DeShon 1998;Huba and Bentler 1982;Huba and Harlow 1987). Methodological advances in structural equation modeling further allow for complex models in which interactions among latent variables can be tested and yield a range of fit indices that simplify the evaluation of model-data fit. ...

Extensive evidence supports associations between early pubertal timing and adolescent externalizing behavior, but how and under which conditions they are linked is not fully understood. In addition, pubertal development is also characterized by variations in the relative speed at which individuals mature, but studies linking pubertal 'tempo' and outcomes are scarce. This study examined the mediating and moderating roles of spare time activities in associations between pubertal development and later delinquency, using data from a large (4,327 girls, 4,250 boys) longitudinal UK cohort (Avon Longitudinal Study of Parents and Children). Self-reports of Tanner stage were available from ages 9 to 14, spare time activities at age 12 and delinquency at age 15. Pubertal development was examined using latent growth models. Spare time activities were categorized using factor analyses, yielding four types (hanging out at home, hanging out outside, consumerist behavior, and sports/games), which were examined as mediators and moderators. Earlier and faster maturation predicted delinquency in boys and girls. Spare time activities partially mediated these links such that early maturing girls more often engaged in hanging out outside, which placed them at greater risk for delinquency. In addition, compared to their later and slower maturing counterparts, boys who matured earlier and faster were less likely to engage in sports/games, a spare time activity type that is linked to lower delinquency risk. No moderation effects were found. The findings extend previous research on outcomes of early maturation and show how spare time activities act as proxies between pubertal development and delinquency.

... Because of the potential problems with using maximum likelihood estimation for data that deviates from multivariate normality, we checked maximum likelihood results with asymptotically distribution free estimates. Results of this analysis and previous research (e.g., Huba & Harlow, 1987) suggest that maximum likelihood estimates are robust. 6 6Asymptotically distribution free estimation provided results that were consistent with the ML results; therefore, they are not reported here. ...

Based on three waves of data from 1261 adolescents, this study examines the nature of resistance self-efficacy vis-a-vis different drugs and social situations, as well as its relationship to perceived pressure to use drugs. We found that both self-efficacy and perceived pressure to use drugs appear to be generalizable across substances (alcohol, cigarettes, and marijuana), but adolescents do tend to distinguish between their capacity to resist drugs in different social situations. Adolescents also discriminate between how much pressure they feel and their ability to resist that pressure, but the great majority report lower levels of self-efficacy in higher pressure situations. This relationship is strongest for alcohol and weakest for marijuana. These results suggest the following implications for prevention programs: (a) adolescents can be taught to resist one or more of the commonly used drugs with a reasonable expectation that the skills will generalize to other drugs; (b) resistance self-efficacy learned in one situation can be expected to have some generalizability to other situations, but it may be important to link resistance training with a range of situations to insure the greatest effectiveness; (c) to be maximally effective, prevention programs may need to help adolescents reduce the amount of pressure experienced as well as develop resistance skills; such efforts are likely to be particularly important for situations involving alcohol.

... To provide a metric for the latent variables (LVs), one of the indicators loading for each variable was set to 1.00. Maximum likelihood estimation was used to generate standardized parameter estimates because it is robust in dealing with deviations from multivariate normality (Huba & Harlow, 1987). ...

The proposition, derived from self-determination theory (SDT), that autonomy-support has a positive effect on self-motivation and well-being, is examined in two distinct cultural settings. Participants were 264 high school students from Russia and the United States who completed measures of perceived parental- and teacher-autonomy-support, academic motivation, and well-being. Means and covariance structure analyses were used to examine the cultural comparability of measured constructs. Results supported the hypotheses that Russian adolescents would perceive parents and teachers as more controlling than U.S. students; and in both samples, perceived autonomy-support would predict greater academic self-motivation and well-being. Results are discussed in terms of SDT’s postulate of a basic human need for autonomy in the context of cultural variations.

... Maximum likelihood estimation was used to generate the standardized parameter estimates because it is robust in dealing with data that deviate from multivariate normality (Huba & Harlow, 1987). To determine the fit of the models to the observed data, we used the chi-square statistic (Bollen, 1989), the nonnormed fit index (NNFI; Tucker & Lewis, 1973), and the root-meansquare error of approximation (RMSEA) (Steiger, 1990). ...

Self-determination theory was applied to explore the motivational basis of adherence to long-term medication prescriptions. Adult outpatients with various diagnoses who had been on a medication for at least 1 month and expected to continue (a) completed questionnaires that assessed their autonomous regulation, other motivation variables, and perceptions of their physicians' support of their autonomy by hearing their concerns and offering choice; (b) provided subjective ratings of their adherence and a 2-day retrospective pill count during an interview with a clinical psychologist; and (c) provided a 14-day prospective pill count during a subsequent, brief telephone survey. LISREL analyses supported the self-determination model for adherence by confirming that patients' autonomous motivation for adherence did mediate the relation between patients' perceptions of their physicians' autonomy support and their own medication adherence. (PsycINFO Database Record (c) 2012 APA, all rights reserved)

... Although causal claims are not warpothetical model tested in this study. Global ranted, the time-ordered nature of the data maternal reports of child behavior prob-makes causal inferences more tenable lems and personal distress predicted proxi- (Huba & Harlow, 1987) and important in a mal reports of the same constructs, as well situation in which experimental manipulaas direct observational measures of dys-tions cannot be set up. However, this is not functional parenting. ...

A community sample of 96 mother-child dyads participated in a study evaluating the extent to which directly observed differences in maternal parenting behavior could be predicted on the basis of both global and proximal maternal reports of child behavior problems and personal distress. To allow for simultaneous testing of a set of relations and make tentative causal inferences, a structural equation modeling approach was used. When the analysis was conducted on the entire sample, results indicated that global and to a lesser extent proximal measures of child behavior problems and personal distress made modest contributions to dysfunctional parenting, with neither child behavior problems or personal distress playing a more important role than the other. When the sample was divided into low (n = 54) and high (n = 42) socioeconomic disadvantage (SED) families, a different picture emerged. In low disadvantage families, parenting was most strongly predicted by mothers' proximal reports of their children's behavior; whereas in high disadvantage families, parenting was best predicted by mothers' proximal reports of their own personal distress. Results are interpreted in light of Wahler and Dumas' (1989) attentional hypothesis. It suggests that mothers who do not experience chronic sources of distress (such as SED) attend and respond to their children's behavior in a fairl accurate and consistent manner, but that mothers who experience chronic distress are unable to attend effectively to their children, responding to them often in light of the many stressors to which they are exposed, rather than in light of the children's actual behavior.

... We used LISREL 8 (Jiireskog and S&-born, 1993) to evaluate each mediation model, analyzing the covariance matrix based on listwise deletion of missing data (Bentler & Chou, 1987; Hayduk, 1987; Bentler & Chou, 1987; Hayduk, 1987). Although presuming multinormality, maximum likelihood estimates are robust against violations of this assumption (Huba & Harlow, 1986, 1987). Unlike traditional path analysis, LISREL estimates various goodness-of-fit indexes (testing parameter restrictions, James, Mulaik & Brett, 1982) and corrects measurement error in parameter estimates. ...

This study evaluated the prevailing explanations for why recruit-ing sources are differentially effective: realism and individual differ-ences. Structural equations modeling analyses tested whether these psychological processes mediate source effects in a sample of 221 nurses. Results supported a model comprising both mediators, but showing that realism processes largely translate how recruiting sources influence job satisfaction, turnover, and absenteeism. However, the study also found direct effects of recruiting source on posthire outcomes, suggesting that unexplored mechanisms may be operating. Implications of these findings are identified and discussed. Employers universally target certain sources from which they recruit qualified personnel for job vacancies (Taylor & Schmidt, 1983). This ubiquitous practice has inspired a myriad of studies during the past 30 years, evaluating how well different recruiting sources generate superior employees (Williams, Labig & Stone, 1993). Early research primarily identified which sources are most effective and what posthire outcomes they produce (Rynes, 1991). By comparison, present-day studies attempt to uncover the reasons for why some recruiting sources are more effective than others (Wanous & Colella, 1989). In particular, scholars have considered two prime explanations why certain recruiting sources outperform others--namely, the individual difference hypothesis and realism hypothesis (Rynes, 1991). The individual difference hypothesis (Schwab, 1982; Taylor 8t Schmidt, 1983) posits that recruiting sources "vary in effectiveness because they are associated with different populations of potential employees" (Wanous & Colella, 1989, p. 8 1). Essentially, recruits from different sources differ in

... Means, standard deviations (Table 1) and intercorrelations were computed for all observed variables (Table 2). Furthermore, values of skewness and kurtosis were computed for all measured variables as the assumption of multivariate normality is a prerequisite to robust SEM (Bollen & Long, 1994;Huba & Harlow, 1987). Cronbach (1951) alpha was computed for examining the internal consistency of the items comprising the theory of reasoned action. ...

Background . Ajzen & Fishbein's (1977) theory of reasoned action and, later Ajzen's planned behaviour theory (Ajzen, 1988; Ajzen & Madden, 1986) have received great attention in the literature (e.g., Giles & Cairns, 1995). A number of researchers have attempted to explain several aspects of human behaviour (e.g., Bandura, 1997). A review of literature between 1990 and 1997 using the Psychlit and ERIC databases which pertained to the application of the theories of reasoned action and/or planned behaviour for the explanation of student study behaviour for an upcoming examination revealed one study (Clarry & Burns, 1991) in which little support for the use of the theory of planned behaviour was provided.
Aims . The purpose of the present study was to examine the appropriateness of the theories of reasoned action (Ajzen & Fishbein, 1977) and planned behaviour (Ajzen & Madden, 1986) to explain student study behaviour during final examinations as pertaining to achieving a high Grade Point Average (GPA).
Sample . This comprised 136 freshmen Greek students from an American undergraduate institution located in Greece.
Method . A structural equation modelling analysis using EQS 4.02 (Bentler, 1992) was used to evaluate the construct validity of the theories of reasoned action and planned behaviour. In this modelling, latent variables were a function of measured variables and direct and indirect relationships were postulated in a path‐analysis framework in order to explain both theories. Since the theory of reasoned action was nested within the theory of planned behaviour, a direct comparison between them was feasible.
Results . Ajzen & Fishbein's (1977) theoretical framework (reasoned action) was well supported, providing a Comparative Fit Index (CFI) of .947. The only modification between the original (Ajzen & Fishbein, 1977) and the final structural model was a direct path between motivation to comply and study behaviour. This structural model provided significant improvement over the original model providing a CFI of .976 and a non‐significant value. Further, inclusion of the perceived control constructs produced a model with acceptable fit also (CFI = .934).
Conclusion . It is concluded that both theories describe well the study behaviour of students although inclusion of perceived behavioural control was not fully supported by the data.

... This is because the sample covariance matrix S is unduly influenced by a small proportion of outliers and can be very inefficient when the sample comes from a distribution with heavy tails (Tyler 1983). To deal with such problems, Jöreskog (1977) and Browne (1982) suggested the possibility of using a robust covariance estimator S n in (1) instead of the sample covariance matrix S. Huba and Harlow (1987) followed up these suggestions, and reported having experimented extensively with the use of resistant covariance matrices in place of ordinary sample covariance matrices in standard structural modeling programs. In spite of these early observations, made one to two decades ago, no technical development has emerged to clarify the precise effects of substituting a robust estimator for S in (1) in the structural equation literature. ...

Existing methods for structural equation modeling involve fitting the ordinary sample covariance matrix by a proposed structural model. Since a sample covariance is easily influenced by a few outlying cases, the standard practice of modeling sample covariances can lead to inefficient estimates as well as inflated fit indices. By giving a proper weight to each individual case, a robust covariance will have a bounded influence function as well as a nonzero breakdown point. These robust properties of the covariance estimators will be carried over to the parameter estimators in the structural model if a technically appropriate procedure is used. We study such a procedure in which robust covariances replace ordinary sample covariances in the context of the Wishart likelihood function. This procedure is easy to implement in practice. Statistical properties of this procedure are investigated. A fit index is given based on sampling from an elliptical distribution. An estimating equation approach is used to develop a variety of robust covariances, and consistent covariances of these robust estimators, needed for standard errors and test statistics, follow from this approach. Examples illustrate the inflated statistics and distorted parameter estimates obtained by using sample covariances when compared with those obtained by using robust covariances. The merits of each method and its relevance to specific types of data are discussed.

... All estimated paths were statistically significant. Using a variety of diagnostic strategies mentioned previously (see, among others, Huba & Bentler, 1982; Huba & Harlow, 1986, 1987), we determined that the misfit was associated with local dependence in two places. First, we allowed risky sex with men, an indicator with an exploratory loading close to our arbitrary |.30| loading criterion , to load on the second general factor as well as the first. ...

This article examines the structure of several HIV risk behaviors in an ethnically and geographically diverse sample of 8,251 clients from 10 innovative demonstration projects intended for adolescents living with, or at risk for, HIV. Exploratory and confirmatory factor analyses identified 2 risk factors for men (sexual intercourse with men and a general risk factor) and 3 factors for women (sexual intercourse with men, substance abuse, and a high risky sex behavior factor). All factors except women engaging in risky sex with men strongly predicted known HIV status of clients for men and women. The findings from this investigation highlight the use of structural equation modeling for applied problems involving overlapping and complex sets of risk behaviors in youth who present at community health programs.

... In such cases, it is better to model a robust covariance matrix S r . Since the empirical experimentations by Huba and Harlow (1987), various technical procedures have been developed for modeling robust covariance matrices (Yuan & Bentler, 1998a, 1998cYuan, Chan, & Bentler, 2000). They differ in how to control the weight attached to each case (see Yuan & Bentler, 1998a, 1998cYuan & Hayashi, 2003). ...

Model evaluation is one of the most important aspects of structural equation modeling (SEM). Many model fit indices have been developed. It is not an exaggeration to say that nearly every publication using the SEM methodology has reported at least one fit index. Most fit indices are defined through test statistics. Studies and interpretation of fit indices commonly assume that the test statistics follow either a central chi-square distribution or a noncentral chi-square distribution. Because few statistics in practice follow a chi-square distribution, we study properties of the commonly used fit indices when dropping the chi-square distribution assumptions. The study identifies two sensible statistics for evaluating fit indices involving degrees of freedom. We also propose linearly approximating the distribution of a fit index/statistic by a known distribution or the distribution of the same fit index/statistic under a set of different conditions. The conditions include the sample size, the distribution of the data as well as the base-statistic. Results indicate that, for commonly used fit indices evaluated at sensible statistics, both the slope and the intercept in the linear relationship change substantially when conditions change. A fit index that changes the least might be due to an artificial factor. Thus, the value of a fit index is not just a measure of model fit but also of other uncontrollable factors. A discussion with conclusions is given on how to properly use fit indices.

... Instead, tetrachoric (dichotomous with dichotomous) polychoric (ordinal with ordinal) 4 and polyserial correlations (continuous with ordinal) should be computed, and the correct asymptotic covariance matrix of such correlations should be analyzed by the method of Weighted Least Squares (WLS), using PRELIS (Jöreskog & Sörbom, 2005a). Failure to do otherwise can lead to gross errors in correlation estimates, distorted parameter estimates, and incorrect goodness-of-fit measures and standard errors (Huba & Harlow, 1987; Jöreskog & Sörbom, 1979, 1988, 1989, 1993). From Jöreskog (1994, p. 383), the special features of ordinal variables are worth noting: ...

... Descriptives. Values of skewness and kurtosis were computed for all variables as the assumption of multivariate normality is a prerequisite to robust SEM (Bollen & Long, 1994;Huba & Harlow, 1987). Cronbach (1951) alpha was computed to examine the internal consistency of the constructs comprising the theory of planned behaviour as it has been found to be a robust statistic (Sideridis, 1999b). ...

The theory of planned behaviour has been rarely used for the explanation of student study behaviour and achievement. Although successful, the theory has been criticised for not including important cognitions, so goal importance was added in the present study. Goal importance refers to the weight-importance an individual assigns towards achieving a specific goal (Hollenbeck & Williams, 1987).
The purpose of Study 1 was to explain the study behaviour habits of first year college students, using a) Ajzen and Madden's (1986) theory of planned behaviour, and b) planned behaviour with the addition of goal importance. The purpose of Study 2 was to replicate the findings of Study 1.
The sample of Study 1 included 149 first year students of an American College located in northern Greece. Study 2 included 85 first year students of the same institution.
The students in Study 1 were given a questionnaire four weeks prior to the end of the spring 1998 semester, and those in Study 2 in the autumn of 1998, including all elements of the theory of planned behaviour and goal importance. The data were modelled using Covariance Structural Modelling (CSM) and EQS 5.7b (Bentler, 1998).
The planned behaviour model was not well supported in Study 1 providing a Comparative Fit Index (CFI) of.83. However, when goal importance was included in the equation, the resulting structural model produced a CFI of.94. The final structural model of Study 1 was re-tested with the sample of Study 2 and produced a CFI =.95.
Findings suggest that goal importance is the causal agent in directing all elements necessary to achieve high levels of study behaviour. Future studies should examine the role of goal importance with other behaviours as well.

On the basis of previous literature (Hays, Widaman, DiMatteo, & Stacey, 1987; Huba, Wingard, & Bentler, 1981), four alternative causal models of substance use were specified, estimated, and evaluated for adequacy of statistical fit for samples of White male (N = 9,164) and female (N = 8,421) adolescents. Results supported the adequacy of a four-variable simplex model, in which alcohol use predicted marijuana use, and marijuana use predicted enhancer and dampener hard drug use. The four-variable simplex model was robust across the male and female adolescent samples. In addition, we found comparisons of maximum likelihood and asymptotically distribution-free estimators to be relatively robust across the causal models specified with respect to the magnitude of the parameter estimates, and with respect to the significance levels of the critical ratios, although chi-square model fit statistics for the maximum likelihood estimates were highly inflated.

How to safely and quickly evacuate before or during a flood is crucial for public safety. In this study, the emergency evacuation capability for rural households to flood hazards was studied using structural equation modeling. Firstly, eleven key factors influencing emergency evacuation capability were identified using existing literature and discussions with some experts and local participants. Secondly, a hypothetical structural equation model on the influencing factors and their relationships to emergency evacuation capability was proposed. Finally, the relationships among influencing factors and their impacts on emergency evacuation capability were analyzed. The results showed that (1) the hypothetical relationships in the structural model were reasonable, and the latent variables in measurement models could be validly measured by observed indices; (2) the emergency evacuation capability could be presented by three latent variables, i.e. personal ability, family characteristics and social environment; and (3) most influencing factors had significant impact on emergency evaluation capability, and the relative importance of the eleven observed indices were information acquiring ability, rapid transfer ability, evacuation skills, early warning, evacuation attitude, the ratio of family members who work and live outside the household, illiteracy ratio, shelter, dependency ratio, flood knowledge, and government assistance. Based on the results, suggestions to improve the emergency evacuation capability for rural households to flood hazards were proposed. The results of this study can be used to improve the emergency evacuation capability for rural households to flood hazards.

The goal of this paper is to compare three estimation methods in analysing a confirmatory factor analysis model using a Monte Carlo simulation design. Nonnormal data are generated and analysed to assess the effects of sample size and estimation methods on parameter bias, standard errors, and chi-square test values.

It was hypothesized that both, adolescents experiencing discrepancies between aspired and attained autonomy, and also adolescents whose parents insist on the keeping to the rules set, would be prone to self-derogation. Contradictory views on control and autonomy were however assumed to be most hazardous to adolescents' self-esteem. Data of 618 participants of the Berlin Longitudinal Study of Youth were analyzed (M = 14.7 years, SD = 0.7) by applying regression models. Taken separately, increased striving for autonomy as well as a high degree of parental strictness could be shown to be related to self-derogation. If both aspects were combined, self-derogation was even more prominent. The results point to the importance of a match between adolescents' aspiration for autonomy and the degrees of freedom granted by their parents.

To assess barriers to implementing standing order protocols (SOP) for vaccinations and influential authorities in making vaccination decisionswith the proportion of black residents, and vaccination coverage in nursing homes. The Centers for Medicare & Medicaid Services and the Centers for Disease Control and Prevention in 2000-2002 surveyed approximately 280 nursing homes in14 states. Data from the On-line Survey and Certification Reporting System were included. A demonstration project to adopt SOPs for vaccination and to assess barriers. Factor analysis and structural equation models were used to assess relationships ofbarriers and influential authorities to implementing SOPs. External facility concerns are barriers to implementing SOPs (p=.031), and nursing homes with higher proportions of black residents are more likely to report those concerns. The medical director and the facility administrator are the most influential authorities determining whether SOPs are implemented. The Quality Improvement Organization (QIO) and the state certification sur-veyor also played important roles in influencing staff making vaccination decisions. The state’s QIO and the state certification surveyor may play important roles in addressingconcerns about staff’s authority to vaccinate under SOPs.Barriers external to the nursing home may play a more important role than internal facility barriers.

This study evaluated the prevailing explanations for why recruiting sources are differentially effective: realism and individual differences. Structural equations modeling analyses tested whether these psychological processes mediate source effects in a sample of 221 nurses. Results supported a model comprising both mediators, but showing that realism processes largely translate how recruiting sources influence job satisfaction, turnover, and absenteeism. However, the study also found direct effects of recruiting source on posthire outcomes, suggesting that unexplored mechanisms may be operating. Implications of these findings are identified and discussed.

This book introduces multiple-latent variable models by utilizing path diagrams to explain the underlying relationships in the models. This approach helps less mathematically inclined students grasp the underlying relationships between path analysis, factor analysis, and structural equation modeling more easily. A few sections of the book make use of elementary matrix algebra. An appendix on the topic is provided for those who need a review. The author maintains an informal style so as to increase the book's accessibility. Notes at the end of each chapter provide some of the more technical details. The book is not tied to a particular computer program, but special attention is paid to LISREL, EQS, AMOS, and Mx. New in the fourth edition of Latent Variable Models: * a data CD that features the correlation and covariance matrices used in the exercises; * new sections on missing data, non-normality, mediation, factorial invariance, and automating the construction of path diagrams; and * reorganization of chapters 3-7 to enhance the flow of the book and its flexibility for teaching. Intended for advanced students and researchers in the areas of social, educational, clinical, industrial, consumer, personality, and developmental psychology, sociology, political science, and marketing, some prior familiarity with correlation and regression is helpful. © 2004 by Lawrence Erlbaum Associates, Inc. All rights reserved.

In the spring of 1990, the Hubble Space Telescope was put into orbit, the culmination of work by a multitude of astronomers, engineers, technicians, and researchers over a period of many years. Its proponents hail it as a key tool to understanding the universe, while its critics write it off as a monumental waste of resources that will never fulfill the expectations of those who designed it. Almost immediately after it went on-line, concern arose about the robustness of its inner workings, yet the demand for access to this device is immense.

The structure of health-related quality of life (HRQOL) in HIV disease is examined in 205 symptomatic HIV+ individuals receiving care at two West Coast public hospitals. A 64-item HRQOL battery, tapping aspects of HRQOL of particular relevance to individuals with HIV disease, was administered and found to yield reliable self-report data. Confirmatory factor analysis provides support for a two-factor model of HRQOL: (a) a physical health dimension defined by physical function, role function, freedom from pain, disability days, and quality of sex life, and (b) a mental health dimension defined by overall quality of life, emotional well-being, hopefulness, lack of loneliness, will to function, quality of family life, quality of friendships, and cognitive function/distress. Correlations of HRQOL measures with social support, access to care, coping, and symptom measures are reported and discussed.

This article proposes a model of sexual activity among secondary school-going Zambian girls. The model is tested for two age categories of adolescents using Covariance Structure Analysis. A sample of teenage girls was selected randomly from seven secondary schools that were selected randomly from schools located in the urban provinces of Copperbelt and Lusaka-Central in Zambia. The article identifies the role of dating as an intervening variable in explaining the variation in sexual activity among teenagers. In Zambia, courtship activities between young men and women traditionally took place under the supervision of parents and community elders. Presently, schools are an important setting for the young to meet and initiate sexual relationships. The theoretical and policy implications of this are discussed.

The increasing popularity of structural equation models that correct for attenuation due to measurement error is noted. The methods by which structural models correct for the effects of measurement error are reviewed. Next, implications of such disattenuation for interpreting the results of structural equation models are considered. Recommendations are advanced for addressing the practice of disattenuation, and caution is urged in drawing inferences based on disattenuated parameter estimates.

A measure assessing an individual's ability to assert the use of condoms was developed using, N= 248, heterosexually active college men and women. Both principal component analysis (PCA) and structural equation modeling (SEM) procedures were performed. External validity for the assertion for condom use measure was established by integrating the measure with the stages of change dimension from the transtheoretical model of behavior change. The transtheoretical model posits that both the cessation of high-risk behaviors and the acquisition of health behaviors involve the progression through five stages of change: precontemplation, contemplation, preparation, action, and maintenance. The results indicated that individuals were further along in the stages of change for using condoms with a casual partner, as compared to a steady partner. The degree to which assertive condom use behavior was engaged in was related to an individual's stage of readiness for using condoms with the two types of partners. The utility of stage-matched intervention strategies, as opposed to the action-oriented approaches to modify high-risk sexual behavior, is discussed.

Data in social science research often have a multilevel structure, with observations on individuals nested within groups. However, such data have been routinely analyzed at the individual level, whereas the group level variation has been ignored. Recent developments in multilevel covariance structure modeling (MCSM; Muthén, 1994) provide a useful statistical tool for analyzing sources of variation due to both within- and between-group influences. The purpose of this study is to illustrate the use of multilevel confirmatory factor analysis (MCFA) via a didactic example of a 2-level (student vs. class) analysis of exercise intrinsic motivation to experience sensations (IM-ES) data. Application of MCFA revealed substantial variation in the IM-ES at both student and class levels. In addition, predictors designed to explain the between-class levels of IM-ES (class size, type of instructor, and type of activity) significantly predicted the class-level variation in the IM-ES. The study demonstrated the merit of using multilevel analytic techniques in substantive research where data have multilevel components.

The structural equations program EQS V3.0 has proved to be an important tool for research in the field of Monte Carlo simulations.
However, when used to replicate a simulation experiment, the simulator has been found to contain a serious flaw which makes
it unusable for particularly complex models. This functional anomaly stems from the fact that, from a given replication onward,
none of the remaining replications offer adequate and convergent solutions.

The purpose of Study 1 was to evaluate how goal-importance (Hollenbeck & Williams, 1987) would fit Ajzen's (1988) theory of planned behaviour in order to explain the study behavior habits of 339 first-year college students who attended an American and a British undergraduate institution. Using Covariance Structural Modeling (CSM) and EQS 5.7b (Bentler, 1998), the planned behaviour model with goal-importance produced a CFI = 0.95. Cross-validation with gender as the factor resulted in good model fit. The purpose of Study 2 was to replicate the findings of Study 1 with a different population, that of high school students. Participants were 773 senior high school students whose study behaviour habits were examined. The final structural model of Study 1 was re-examined with this new sample and produced a CFI = 0.97. Cross-validation across gender again produced good model fit. It is concluded that there is ample evidence that goal-importance is the causal agent in directing all elements necessary to achieve high levels of study behaviour given the theoretical framework of planned behaviour. More research is needed, however, to examine its role with other behaviours (e.g. procrastination) or academic achievement as well.

Testing fit is a critical and controversial step in structural equation modeling (SEM), and alternative fit indices have been proposed. To provide guidance, a simulation study was conducted to evaluate the effects of estimation method, number of indicators per factor (p/r ratio), sample size, and loading size on six SEM fit indices: chi-square/degree of freedom ratio (chi(2)/df), Normed Fit Index (NFI), Nonnormed Fit Index (NNFI), Centrality m index, Relative Noncentrality Index (RNI), and Comparative Fit Index (CFI). When improper solutions occurred, the effects of constraining out-of-bounds estimates to a reasonable range (the method used in EQS software) on the fit indices was evaluated. Four levels of sample size (50, 100, 200, and 500), three levels of loading size (0.50, 0.70, and 0.90), five levels of p/r ratio (2, 3, 4, 5, and 6), and two levels of estimation method (Generalized Least Squares [GLS] and Maximum Likelihood [ML]) were examined. The results of this study indicated that: (a) improper solutions occurred frequently when p/r = 2 and sample size was small or loadings were low; (b) GLS produced more improper solutions than ML in general; (c) there was no effect of improper solutions (constrained to some boundary) found for any of the fit indices but NFI (more downward bias occurred for NFI when improper solutions were present); (d) four incremental fit indices (NFI, NNFI, RNI, and CFI) were negatively affected by increasing the p/r ratio; (e) NFI was affected much more seriously than the other three fit indices and should not be used; (f) all fit indices except NNFI were found to be significantly affected by estimation method (less bias occurred for GLS than for ML); and (g) interaction effects between estimation method, p/r ratio, sample size, and loading size also occurred. Recommendations regarding selection of a fit index are made based on the findings.

Two different approaches are used to assess the one-month stability of depressive affect in college students. First, a high retest correlation is demonstrated for a latent depressive affect construct using self-reports from the previous month on the Beck Depression Inventory, the Center for Epidemiological Studies-Depression Scale, and Zung's Self-Rating Depression Scale. This is true although the corresponding stabilities for individual scales are only moderate. Second, the predictive validity of depressive categorization based on these scales is examined using logistic regression techniques. Depressive categorizations are best predicted from the corresponding depressive categorization on the same scale at Time 1. There is little additional across-scale predictive information in the assessment of depressive categorization; this again suggests that a single factor model is most appropriate. Further, the association between assessments of dysphoric affect over one month in college students is less than previously established longitudinal associations between dysphoric affect categorizations over one year in older adult samples. This suggests that college students may be more affectively volatile than adults.

It is commonly thought that structural equation modeling corrects estimated relationships among latent variables for the biasing effects of measurement error. The purpose of this article is to review the manner in which structural equation models control for measurement error and to demonstrate the conditions in which structural equation models do and do not correct for unreliability. Generalizability theory is used to demonstrate that there are multiple sources of error in most measurement systems and that applications of structural equation modeling rarely account for more than a single source of error. As a result, the parameter estimates in a structural equation model may be severely biased by unassessed sources of measurement error. Recommendations for modeling multiple sources of error in structural equation models are provided. (PsycINFO Database Record (c) 2012 APA, all rights reserved)

[Correction Notice: An erratum for this article was reported in Vol 75(1) of
Journal of Applied Psychology (see record
2008-10492-001). An error exists in Figure 2 and the accompanying text of the article. The corrected information is included in the erratum.] The problem of assessing fit of structural equation models is reviewed, and two sampling studies are reported that examine the effects of sample size, estimation method, and model misspecification on fit indices. In the first study, the behavior of indices in a known-population confirmatory factor analysis model is considered. In the second study, the same problem in an empirical data set is examined by looking at antecedents and consequences of work motivation. The findings across the two studies suggest that (a) as might be expected, sample size is an important determinant in assessing model fit; (b) estimator-specific, as opposed to estimator-general, fit indices provide more accurate indications of model fit; and (c) the studied fit indices are differentially sensitive to model misspecification. Some recommendations for the use of structural equation model fit indices are given. (PsycINFO Database Record (c) 2012 APA, all rights reserved)

Addresses common questions about the methodology of structural equation modeling, which is appropriately used whenever models of relationships among variables need to be estimated and tested. (PsycINFO Database Record (c) 2012 APA, all rights reserved)

investigated the potential processes through which one class of deviance may influence another type, by evaluating the longitudinal effects of a specific form of deviance (drug use) in adolescence on a different form of deviance (criminal behavior) in adulthood / the prospective effects of constructs relevant to strain, control, and social learning theories were also studied / examines 2 waves of data within [a] larger study that provided assessments of the [theoretical] contructs of interest over an 8-yr interval of time, from late adolescence to adulthood / the mean age of the Ss at the 1st time of measurement used in this study was 18.9, and was 26.9 . . . at the 2nd time of measurement
effects studied in terms of traditional perspectives are described first, to provide a theoretical background helpful in [the] subsequent outline of alternative explanations of associations between drug use and criminal deviance (PsycINFO Database Record (c) 2012 APA, all rights reserved)

In this chapter we will present a tutorial on the fundamentals of structural equation modeling (SEM). In recent years structural equation modeling (SEM) has grown enormously in popularity. Fundamentally, SEM is a term for a large set of techniques based on the general linear model. The goal of this chapter is to present an introduction to the field of structural equation modeling. The chapter begins with an overview of SEM that includes the purpose and goals of this statistical analysis as well as terminology unique to this technique. Following the brief overview, the process of modeling will be discussed and illustrated with an example using data from a recent D.A.R.E. evaluation. After the basic modeling process is illustrated and discussed, an extension (multilevel modeling) of the basic model is presented. Finally, we present a section that discusses future trends in SEM.

Agnew's (1985) revised strain theory argues that delinquency results from the inability to escape legally from painful or aversive situations. There is much indirect support for the theory, with experimental and survey data indicating that delinquency is associated with a wide variety of aversive situations. The experimental data, however, are of questionable generalizability and the survey data are primarily cross sectional in nature, leaving unresolved the issue of causal direction. This study examines the relationship between environmental adversity and delinquency using longitudinal data from a nationally representative sample of adolescent boys. A distribution-free method (arbitrary generalized least squares) is used to estimate a covariance structure model depicting a reciprocal relationship between adversity and delinquency. The results support the revised strain theory: environmental adversity has a causal effect on delinquency, but delinquency does not have an effect on adversity.

Methods for obtaining tests of fit of structural models for covariance matrices and estimator standard errors which are asymptotically distribution free are derived. Modifications to standard normal theory tests and standard errors which make them applicable to the wider class of elliptical distributions are provided. A random sampling experiment to investigate some of the proposed methods is described.