EFFECTS OF JOB-RELATED STRESS ON FACULTY INTENTION TO LEAVE ACADEMIA

The purpose of this study was to investigate the relationship between job-related stress and faculty intent to leave academia. The stress variables studied were reward satisfaction, institutional/departmental reputation, time commitment, departmental/institutional influence, and student interaction. We hypothesized that the relationship between these variables and faculty intent to leave academia would be moderated by interest in one's discipline and sense of community — an institutional fit variable. We also investigated the effects of academic discipline, tenure status, and gender on these relationships. Based on data from a national faculty survey of 3,070 full-time tenure-track faculty, results indicated that of the variables studied, the two major correlates of intent to leave academia were time commitment and sense of community; however, time commitment did not moderate the stressor-intent relationship. Though showing significant zero-order correlations with intent, when gender and tenure status were added to the hierarchical regression analyses containing the stressors and moderators, neither variable contributed meaningfully to the prediction of intent. Academic discipline classification (Biglan, 1973) contributed only 2% to explained variance. A prediction model that contained all stressors, both moderators, and the background variables of gender and academic discipline accounted for 25% of the variance in intent to leave academia.     

The Role of Centering for Interaction of Level 1 Variables in Multilevel Structural Equation Models

This article examined the role of centering in estimating interaction effects in multilevel structural equation models. Interactions are typically represented by product term of 2 variables that are hypothesized to interact. In multilevel structural equation modeling (MSEM), the product term involving Level 1 variables is decomposed into within-cluster and between-cluster random components. The choice of centering affects the decomposition of the product term, and therefore affects the sample variance and covariance associated with the product term used in the maximum likelihood fitting function. The simulation study showed that for an interaction between a Level 1 variable and a Level 2 variable, the product term of uncentered variables or the product term of grand mean centered variables produced unbiased estimates in both Level 1 and Level 2 models. The product term of cluster mean centered variables produced biased estimates in the Level 1 model. For an interaction between 2 Level 1 variables, the product term of cluster mean centered variables produced unbiased estimates in the Level 1 model, whereas the product term of grand mean centered variables produced unbiased estimates for the Level 1 model. Recommendations for researchers who wish to estimate interactions in MSEM are provided.     

Investigating efficacy expectancy as criterion for comparison of teacher- versus student-regulated learning in higher education

In this study, we assessed the feasibility of a specific elaboration of the efficacy construct, distinguishing between personal and contextual aspects, as a criterion for comparing learning environments. The participants were 163 students from two student-regulated and two teacher-regulated programs in higher education. We measured students’ perceptions of autonomy and various aspects of perceptions of efficacy in common learning tasks. Using principal components analyses, we assessed the structure of all the relevant variables. Subsequently, analyses of variance were performed with regulation source, discipline and grade level as factors. All variables emerged as separate scales with high internal consistency. Students in student-regulated programs reported a higher level of organisational efficacy expectancy, implying that these students perceived better organisational conditions for supporting their task performance. It is concluded that the distinction between personal and contextual aspects of efficacy is a promising distinction. However, a more fine-grained conceptualisation of teacher and student regulation is needed.     

A Structural Equation Modeling Evaluation of the General Model of Instructional Communication

The General Model of Instructional Communication introduced by McCroskey, Valencic, and Richmond (2004) is supported in its original conception by canonical data. This study, however, uses structural equation modeling (SEM) to provide a more detailed analysis. Although the model as originally hypothesized fits the data poorly, analysis of the SEM results suggests adjustments to the original model that substantially improve the model's fit. The revised model accounts for significant portions of the variance in the outcome variables, provides a more detailed explanation of the relationships involved, and has implications for future research. Bootstrapped parameter estimates suggest that the results are replicable.     

Predictors of parents' physical disciplinary practices.

OBJECTIVE: This study examined how childhood history of discipline (1) related to ratings of how severe and typical punishments were; and (2) predicted parents' use of discipline techniques. The influence of child culpability on these ratings was also investigated. METHOD: Ninety-nine New Zealand parents rated 12 physical discipline scenarios varying in discipline severity and perceived child culpability. Parents judged how severe and typical they considered the disciplines depicted in the scenarios and reported on how often they had experienced such discipline as children and how often they had used them with their own children. RESULTS: When the child was perceived to be at fault, parents rated the discipline depicted as less severe, considered the technique more typical, reported they had been similarly disciplined more frequently, and applied such discipline to their child more frequently. Childhood history of a discipline was related to the parent's use of that method, and the parents judged techniques they used with their own children as less severe and more typical of methods of discipline. History of discipline and severity judgments were the best predictors of parents' disciplinary practices. CONCLUSIONS: Although the findings support the cycle of violence theory, more complex potential pathways to abusive parenting, including the variables in this study, are proposed.     

Student-focused approaches to teaching in relation to context and teacher characteristics

The purpose of this paper is to gain more insight into the relationship between teachers’ approaches to teaching on the one hand, and the characteristics of context and teacher demographics on the other. Data were collected from 50 teaching staff at the University of Antwerp and from three sources: a Dutch translation of the Approaches to Teaching Inventory (ATI), information given by the participants, and information obtained through the personnel department of the university. Only the conceptual change/student-focused scale of the ATI had good reliability and was used for further analysis. Analysis of variance (ANOVA) showed no relationship between teachers’ approaches to teaching and the context variables of expert level of students, teaching discipline and the number of students in the classroom. Neither was a relationship found between the teachers’ conceptual change/student-focused approach and the teacher characteristics of gender, academic status, teaching experience, age and intention to participate in teacher training. Several interpretations of these data and perspectives for further research are discussed.     

Risk and protective factors for poor school adjustment in lesbian,gay, and bisexual (LGB) high school youth: Variable and person‐centered analyses

This study examined the relations between school climate and school adjustment among 101 lesbian, gay, and bisexual (LGB) high school students and the moderating influence of social support on those relations. Students completed surveys to assess three aspects of the school climate (the school's exclusion/inclusion of LGB people, personal victimization in school for being LGB, and social support from teachers) as well as social support from family and close friends. Criterion variables were GPA, school belonging, and discipline problems. School climate variables explained significant amounts of variance in all criterion variables, after controlling for prior GPA, and there were no moderating effects of parent or friend support. Cluster analyses revealed one small group (n = 14) of highly vulnerable youth who were the least adjusted, most victimized, and least supported. Implications for teachers, counselors, and future research are discussed. © 2005 Wiley Periodicals, Inc. Psychol Schs 42: 159–172, 2005.     

Problem length as a structural variable in verbal arithmetic problems

Of five structural variables that in previous studies accounted for a significant amount of the observed variance in the error rate in verbal arthmetic problems, the length variable (number of words in the problem statement) apparently was more important in the upper grades than in the lower grades. In this study three forms of a verbal problem set in which the number of words in the problem statements were systematically varied were administered to classes of students in Grades 4–9. Using regression analysis, the investigator found that although the length variable did not measure a significant amount of variance between two of the forms, with the third it did.     

The Color of Discipline: Sources of Racial and Gender Disproportionality in School Punishment

The disproportionate discipline of African-American students has been extensively documented; yet the reasons for those disparities are less well understood. Drawing upon one year of middle-school disciplinary data for an urban school district, we explored three of the most commonly offered hypotheses for disproportionate discipline based on gender, race, and socioeconomic status. Racial and gender disparities in office referrals, suspensions, and expulsions were somewhat more robust than socioeconomic differences. Both racial and gender differences remained when controlling for socioeconomic status. Finally, although evidence emerged that boys engage more frequently in a broad range of disruptive behavior, there were no similar findings for race. Rather, there appeared to be a differential pattern of treatment, originating at the classroom level, wherein African-American students are referred to the office for infractions that are more subjective in interpretation. Implications for teacher training and structural reform are explored.     

Causes and consequences of student evaluation of instruction

Student evaluation of course quality was examined as an intervening variable in a path model with seven input variables and the following two output variables: (1) whether students would recommend the course to others, and (2) students' plans to take more courses in the discipline. Student evaluation of the lecturer was the most powerful input variable for the intervening variable course rating. Class format, defined as lecture or lecture with small groups, did not affect course ratingper se but did relate to students' attitudes toward taking more courses in the discipline. Lecturer and textbook evaluation had significant independent effects on whether the student would recommend the course to others. The major implication for university departments is that quality of instruction in introductory courses has impact on future enrollment plans.     

A causal model of students' achievement in a college physics course

To explain students' achievement in the mechanics segment of a college physics course, a causal model is proposed whose variables are the three constructs: Newtonian physics, math ability, and science experience. Each variable specifies several variates, whose numerical values were measured and intercorrelations computed. The Factorial Modeling (FaM) procedure is applied to the data to estimate the coefficients of the proposed model's structural equations. The FaM results indicate that the model is plausible and that its first two variables account for approximately 34% of the variance in the criterion variate, students' mechanics achievement. The implications of the findings for physics instruction are discussed.     

Unpacking teachers’ intentions to integrate technology: A meta-analysis

The Technology Acceptance Model (TAM) is a key model describing teachers' intentions to use technology. This meta-analysis clarifies some of the contradictory findings surrounding the relations within the TAM for a sample of 45 studies comprising 300 correlations. We evaluate the overall fit of the TAM and its structural parameters, and quantify the between-sample variation through meta-analytic structural equation modeling. The TAM fitted the data well, and all structural parameters were statistically significant. On average, the TAM variables explained 39.2% of the variance in teachers' intentions to use technology. Several sample, measurement, and publication characteristics, including teachers’ experience and the representation of the TAM variables, moderated the relations within the TAM. Overall, the TAM represents a valid model explaining technology acceptance—however, the degree of explanation and the relative importance of predictors vary across samples. Implications for further research, in particular the generalizability of the TAM, are discussed.     

An Evaluation of Normal School Sororities

This investigation’s purpose was to answer the following question: In an in-school suspension setting, does a program of therapeutic discipline that involves counseling, bibliotherapy, writing therapy, and contingency contracting result in more positive attitudes toward school attendance, improved attendance, and greater insight into attendance problems among adolescent truants who participate in the program than among adolescent truants who participate in a traditional program of non-therapeutic discipline? When a one-way analysis of variance was used to analyze differences for attitudes and insights and analysis of covariance was used to analyze behavioral differences among 148 subjects in the two programs, it was found that therapeutic discipline students (a) held less positive attitudes toward school attendance, (b) had better records of attendance in classes and fewer truant absences, and (c) demonstrated greater insight into attendance problems than did non-therapeutic discipline students.     

Why Do Achievement Measures Underpredict Female Academic Performance?

In the context of college admissions, the current study examined whether differential prediction of first‐year grade point average (FYGPA) by gender could be explained by an omitted variable problem—namely, academic discipline, or the amount of effort a student puts into schoolwork and the degree to which a student sees him/herself as hardworking and conscientious. Based on nearly 10,000 college students, the current study found that differences in intercepts by gender were reduced by 45% with the inclusion of academic discipline in a model that already included high school grade point average (HSGPA) and ACT Composite score. Moreover, academic discipline resulted in an additional 4% of variance accounted for in FYGPA. Gender differences in slopes were not statistically significant (p > .001) regardless if academic discipline was included in the model. The findings highlight the utility of taking a more holistic approach when making college admission decisions. Namely, the inclusion of noncognitive measures has benefits that are twofold: increased predictive validity and reduced differential prediction.     

The changing academic environment and diversity in students' study philosophy,beliefs and attitudes in higher education

Student populations in higher education in Australia and elsewhere in the developed world have experienced significant diversity over the past two decades. The existing literature has provided limited clarity about the effects of this diversity on the dimensions underpinning students' study philosophy domain. Based on a large data set from a leading Australian university, this paper analyses students' study philosophy, beliefs and attitudes towards teaching and learning. Factor analysis explored themes (or dimensions) within the survey. Multivariate analysis of variance used these dimensions as dependent variables with age, sex, ethnicity, study discipline, study level, academic performance and sex/ethnicity interaction as grouping variables to identify significant sources of variations. Deep learning, expediency and responsibility reflected the students' study philosophy domain. Deep learning and responsibility varied with ethnicity and academic performance. Expediency differed according to ethnicity, study discipline and academic performance. Students in business-related disciplines displayed greater expediency than peers elsewhere, treating education like any other commodity. The contribution of this study lies in its rigorous analysis of the impact of the diversity of the student population on the study philosophy domain, compared to the existing literature.     

Problem length as a structural variable in verbal arithmetic problems

Of five structural variables that in previous studies accounted for a significant amount of the observed variance in the error rate in verbal arithmetic problems, the length variable (number of words in the problem statement) apparently was more important in the upper grades than in the lower grades. In this study three forms of a verbal problem set in which the number of words in the problem statements were systematically varied were administered to classes of students in Grades 4–9. Using regression analysis, the investigator found that although the length variable did not measure a significant amount of variance between two of the forms, with the third it did.This study was begun while the author was a staff member of The Institute for Mathematical Studies in the Social Sciences at Stanford University. The research was partially supported by National Science Foundation grant G18709.     

Problem length as a structural variable in verbal arithmetic problems

Of five structural variables that in previous studies accounted for a significant amount of the observed variance in the error rate in verbal arthmetic problems, the length variable (number of words in the problem statement) apparently was more important in the upper grades than in the lower grades. In this study three forms of a verbal problem set in which the number of words in the problem statements were systematically varied were administered to classes of students in Grades 4–9. Using regression analysis, the investigator found that although the length variable did not measure a significant amount of variance between two of the forms, with the third it did.This study was begun while the author was a staff member of the Institute for Mathematical Studies in the Social Sciences at Stanford University. The research was partially supported by National Science Foundation grant G18709.     

How Fair are Student Ratings in Assessing the Teaching Performance of University Teachers?

This paper examines the effects of course characteristics on student ratings of university teaching. The multivariate effects of various course characteristics on student ratings of university teachers were examined by a 5×4×3×2×2 multivariate analysis of variance (MANOVA) with academic discipline, class size, level of course, type of course, and mode of study as independent variables and a set of six composite student ratings as the dependent variable. Data were obtained from a university in Hong Kong and consisted of 4996 course average ratings collected from 25 departments over two academic years. Significant differences were found in course average student ratings across academic disciplines, class size groups, course levels, types of course, and modes of study. The effect sizes of the largest differences between sub‐groups within each of the factors were also estimated. Academic discipline differences and class size were found to have the largest effect on student ratings, with effect sizes of 0.43–0.86 and 0.55–1.01 respectively. Implications of the findings are discussed.     

Peer review and panel decisions in the assessment of Australian Research Council project grant applicants: what counts in a highly competitive context?

Applications submitted to the Australian Research Council Large Grants Scheme in five discipline groups were examined with a view to determining the extent of influence of biographical and academic track record variables on ratings by independent assessors (peers) and on final outcomes as determined by the Council's discipline review panels. Variables considered included age, gender, type and status of position, institutional base, previous grants history and publication records of the applicants. Age and gender of the applicants did not have a significant impact on outcomes. Those in research only positions were rated more highly, and those in more prestigious institutions were more likely to win support, but these factors were partly accounted for by the more impressive publication records of these groups. The academic status of the applicant was, however, found to make a significant contribution to an explanation of the variance in independent assessor's ratings of the strength of the researcher, along with but additionally to that contributed by publication components of their track record. Assessor's ratings of the strength of the proposal were, as would be expected, the major influence on the final decisions made by review panels, although academic status of the applicant did, again, make a small but significant additional contribution to an explanation of variance in outcome. These results lend some support to the idea that the Matthew effect, or theory of accumulative advantage, has some impact on peer and panel review of applications within the highly competitive arena of Australian Research Council large grants.