Search

The methods used to analyze the status of women academics at Queen's University, Ontario, Canada, are described. An initial study examined several individual variables believed to affect salary and compared the findings for men and women faculty and utilized a multiple regression analysis to integrate most of those variables into one model. A second study of the same faculty used a simple statistical review similar to that used in the initial study and an adaptation of a peer pairing approach. This study focused on those faculties and discipline groups where most women held positions. Patterns of appointment and promotion, as well as salary review over a 10-year period, are discussed. Both the dean and the individual woman faculty member were asked to name a peer, since the choice of peer reflects a judgment about where an individual is or ought to be. The peers and the women were asked to complete detailed questionnaires on qualifications, experience, and current activities. Comparative salary levels and salary histories were also obtained on the peer pairs. The limitations of quantitative methods and of the peer pairing method are discussed. Integration of traditional review procedures with techniques of statistical analysis and peer pairing allows greater equity at both the group and individual level than any of these methods separately. (SC)

A pilot project in performance funding was conducted at Tennessee Technological University. A random sample of 112 seniors participated in a special assessment exercise involving the American College Testing (ACT) Battery and the ACT College Outcome Measures Project. The participants also completed a questionnaire designed to elicit self-reports of progress toward the realization of a set of institution-wide or general education goals. Factor analyses and item analyses were performed on self-report responses and two separate two-item indices were obtained. The relationships between the "objective," or test performance, and the "subjective," or student testimony, instructional outcomes measures were examined. The findings support the concurrent validity of the student testimony data, but the magnitude of the correlation coefficients seem to suggest that a substantial proportion of the variance in self-reported progress scores is not explained by the test performance results. The implications of the study results for the better-informed selection, implementation, and interpretation of instructional outcomes measures are discussed. Statistical data and a bibliography are included. (SW)

The development of a predictive equation capable of explaining a significant percentage of enrollment variability at Florida International University is described. A model utilizing trend analysis and a multiple regression approach to enrollment forecasting was adapted to investigate enrollment dynamics at the university. Four independent variables were used in a stepwise linear regression analysis to explain variance in the total university full-time-equivalent enrollment. Three of the four variables used are presently employed in business forecasting models to account for changes caused by trends, seasonal variations, and cyclical variations. The fourth variable was introduced to account for the opening and operation of a new campus. The model accounts for 99.6 percent of the variation observed in total university enrollments for the past 20 quarters and provides estimates of the enrollment within one percent of the enrollments actually observed. The model also has the potential for making reasonably accurate and reliable short-range enrollment forecasts. The model demonstrated that changes in economic conditions in the county, reflected by changing unemployment rates, correlate significantly and positively with changes in total university enrollment. Total university enrollment did not change from its second to its sixth year of operation except for those changes that could be accounted for by changes in unemployment in the county. The opening and operation of the new campus increased total university enrollment by about six percent. (SC)

A search for better predictors of incomes of high school and college graduates is described. The accuracy of the prediction, implicit in the work of John R. Walsh of Harvard University, that the income differences in a given year are good indicators of income differences in future years, was tested by applying standard statistical procedures to data describing the annual incomes of men 25 years old and older during the period 1956 to 1975. Addition of the gross national product in the year prior to the previous equation yields much improved results. Application of Richard B. Freeman's three-equation recursive adjustment model indicates that the starting salaries of college graduates are not good correlates of income differences. Several sets of data from the American Council on Education "Fact Book on Higher Education" were then entered into a series of multiple correlation and regression studies. Studies of the 184 independent variables identified four that, taken together, yield significantly better predictions of future income differences. These factors were: (1) male population 22 years old, 13 years prior; (2) male college graduates at the B.A. and first-professional degree level/male population 29 years old, 11 years prior; (3) male college graduates at the B.A. and first-professional degree level, 14 years prior; and (4) male college graduates at the B.A. and first-professional degree level/male population 29 years old, 14 years prior. However, no theory to explain the connections among these variables is apparent. (SC)

Regression formulations and related concerns involved in dealing with wage discrimination suits involving college faculty are considered. Several models based in hypothesis testing "policy capturing" uses of multiple linear regression are presented, and problems related to the court's view of statistical significance are examined. Multiple linear regression can be used to determine if there is a policy operating in the hiring process, and if so, to capture it and evaluate it in terms of R2 (R to the power of 2). The process is to add variables to the model explaining the decision to hire or not to hire until a high level of R2 is achieved, evaluate the weighting coefficients to determine the emphasis given each variable, and then test for significance using the F test, which is described. In the event that the plaintiff has introduced a regression model that shows gender to be a statistically significant contributor to salary and particularly if the institution is large, so that virtually any element related to salary will have some degree of statistical significance, a useful strategy is to introduce similar models based on more appropriate assumptions about the group to be studied. Regressions by rank or based on members of only those departments that have both males and females are alternative approaches. Another approach is to study the effect that gender has had on salaries over some interval of time. A repeated measures model can help determine whether corrective measures reduced differences in salary that can be attributed to gender for individuals who were in the system over a given period of years. The issues of statistical versus practical significance, denominator degrees of freedom, and future use of a path analytic approach are considered. (SW)

Technical considerations in the development of a salary equity model based upon regression analysis are reviewed, and a simplified salary prediction equation is examined. Application and communication of the results of the analysis within the existing operational context of a postsecondary institution are also addressed. The literature is reviewed, with an example of the application of the model provided. The model to predict salaries incorporates in the prediction equation rank, highest degress, administrative status, discipline, years-in-rank, endowed chair, and contract status. The total university prediction equation does not include discriminatory variables such as sex, race, age, or years-employed. A separate equation, incorporating these biased variables, which tests for the possible presence of salary inequities, is discussed. All instructional faculty on a contractual basis, numbering between 500 and 1,000, are included in the regression equation. Results from a multiple regression equation model guide the identification of individual faculty members who appear to be making high or low salaries for their status, discipline, and experience. A high level of predictability was shown in the model, and this result was in conformance to the actual salary structure. It is suggested that fundamental political skills are essential in balancing the forces that affect salaries, but a salary equity study can provide comprehensive information that decision-makers value when they manage conflict. A bibliography is appended. (SW)

Modern technologies radically simplify the availability of the latest scientific literature and offer new possibilities for sharing knowledge. Yet, most higher education institutions still rely on traditional face-to-face teaching and use e-courses "only" to supplement it. Such a combination of teaching methods is known as blended learning and is also used at the Faculty of Administration (FA), a member of the University of Ljubljana. The paper presents the results of a survey among FA students. The study's aim was to identify which aspects of blended learning increase students' knowledge level. Students evaluated 23 different aspects of blended learning on a 7-level scale. As a measure of the level of students' knowledge gained from each course, we used their final grades. Applying principal component analysis, we extracted six dimensions of blended learning which represented predictor components in a multiple linear regression with final grade as the dependent variable. Since courses at FA vary in many aspects, we performed the regression analysis for each obligatory course individually. Our analysis revealed for which courses the final grade can be reliably predicted from the aspects of blended learning. The study also showed that the aspects of traditional face-to-face learning are more strongly linked to better grades than the aspects of e-courses. However, for some courses certain characteristics of e-courses play a significant role in the final grade received. [For the complete proceedings, see ED579335.]

The purpose of this paper is to provide guidance in choice of analytic bias reduction methods for educational studies in which the goal is to estimate a treatment effect in the presence of selection bias into treatment. In addition, issues of dimensionality, collinearity, omitted confounders, missing outcomes, and non-independence may be factors influencing analytic choice. The present paper investigates these issues by systematically comparing performance of common bias reduction techniques, including multiple regression/analysis of covariance (direct control of group differences on covariates), propensity score (PS) matching methods (1:1 or 1:n), and inverse probability weighting on the PS (IPW). Preliminary simulation results showed that MR may be preferred for situations without dimensionality or collinearity issues; otherwise IPW may be preferred.

Findings of a paper that compared three analytic models for classifying public schools are presented in this paper. The first model is the traditional additive multiple linear regression model that assumes the use of a compensatory policy in assigning weights to school attributes. The second is an adaptation of Coomb's (1964) conjunctive model proposed by Einhorn (1970), and the third is an adaptation of Coomb's disjunctive model, also formulated by Einhorn (1970). Data were collected through a "policy capturing" procedure, in which 28 educational leaders in a southeastern state reviewed a series of hypothetical school profiles of 80 elementary and 80 secondary schools and classified them as nonstandard (deficient), standard (acceptable), or exemplary (normatively outstanding). An index of intrajudge consistency was developed to compare the three models of school classification. Multiple regression analysis was used to examine statistical differences. Findings indicate that the distributions of intrajudge consistency indices were largely overlapping. However, the conjunctive model for aggregating school scores was superior in that it clearly differentiated individual school characteristics and provided a more accurate overall picture of school quality. Five tables and three figures are included. (Contains 62 references.) (LMI)

Studies comparing virtual and physical manipulative environments (VME and PME) in inquiry-based science learning have mostly focused on students' learning outcomes but not on the actual processes they engage in during the learning activities. In this paper, we examined experimentation strategies in an inquiry activity and their relation to conceptual learning outcomes. We assigned college students to either use VME or PME for a goal directed physics inquiry task on mass-spring systems. Our analysis showed that the best predictors of learning outcomes were experimental manipulations that followed a control of variable (CV) strategy, with a delay between manipulations ("systematic inquiry"). Cluster analysis of the prevalence of these manipulations per participant revealed two distinct clusters of participants, systematic inquiry or not. The systematic inquiry cluster had significantly higher learning outcomes than the less systematic one. Furthermore, the majority of the participants using the PME belonged to the more systematic cluster, while most of the participants using the VME fell into the non-systematic cluster, likely because of the specific affordances of the real and virtual equipment they were using. However, beyond this impact on inquiry process, condition had little effect. In light of these results, we argue that investigating processes displayed during learning activities, in addition to outcomes, enables us to properly evaluate the strengths and weaknesses of different learning environments for inquiry-based learning. [For complete proceedings, see ED560503.]

Testing learners' English proficiency is central to university English classes in Japan. This study developed and implemented a set of parallel online receptive aural and visual vocabulary tests that would predict learners' English proficiency. The tests shared the same target words and choices--the main difference was the presentation of the target words. Test results showed that for both the Aural Test (AT) and the Visual Test (VT), average scores from 1,000 to 4,000 of the JACET8000 word list's frequency bands decreased in proportion to word frequency. More detailed analyses highlighted gaps between visual and aural aspects of receptive vocabulary knowledge of frequency of 4,000 and beyond. It further indicated that the aural version of the test was a better predictor of English proficiency. For future directions, we need to develop other sets of the VT and the AT to improve the reliability of the tests, and this is especially important in order to satisfy the needs of educational institutes that demand high accuracy in vocabulary testing. [For the complete volume, see ED578177.]

Many technologies have changed the way individuals live and learn. Google Inc. has played significant roles in business and academic worlds. Google Apps for Education and Google Classroom have been offered to higher institutions around the globe. Although large cloud service provider such as Google do not encrypt all their stored electronic data and correlate identifiable data across accounts, Google Drive has been a popular feature of Google for teachers and students in Thai higher educational institutions. The purpose of this research is to explore the impact of message quality and digital literacy in terms of technological dimensions toward intention to use of Google Drive of Bangkok University's students. However, the four dimensions were extracted from fourteen statements. The factors were message quality, digital literacy in terms of learning new technology, digital literacy in terms of technical and IT skills, and intention to use. Furthermore, message quality, digital literacy in terms of learning new technology, and digital literacy in terms of technical and IT skills positively affected intended usage of the system. [For full proceedings, see ED571332.]

This study explores the impact of motivational goals for using social networking sites (SNS) usage and computer self-efficacy towards e-learning effectiveness of the 155 students from different schools at a private university, in Pathum Thani province, Thailand during April to May of academic year 2015/2016. Social dimension and human interaction have played more significant roles in learning and teaching especially in higher institutions. Web 2.0 introduced many users to generate, share, and reuse contents using SNS like Facebook, Twitter, and LINE. LINE chat application was the second largest SNS application utilized by Thai users. Thai students, who often prefer to listen rather than speak to communicate with their teachers, feel more comfortable to chat with their teachers through virtual worlds using LINE. The results revealed that there was positive influence of motivational goals for using SNS usage and computer self-efficacy towards e-learning effectiveness. The highest impact was motivational goals for using SNS usage followed by computer self-efficacy affecting e-learning effectiveness. Possible avenues for future research are also suggested in this study. [For full proceedings, see ED571332.]

Apart from Internet addiction, smartphone addiction has played important roles in students' lives as observed in recent studies. There are positive and negative aspects in using smartphones especially in higher education. University administrators or instructors should take into account what factors are affecting students' smartphone addiction in order to limit its negative effect and introduce more innovative measures such as developing ways of teaching and learning in digital age using smartphones. The aim of this study is to investigate the impact of self-regulation and compulsivity towards smartphone addiction of university students in two private universities in Thailand. 157 survey questionnaires were collected during April to May 2016 and analyzed with multiple regression analysis. Although the conceptual model explains 60.40% of the variance in smartphone addiction, only compulsivity is found to have significant determinant of smartphone addiction. Self-regulation has no significant effect on smartphone addiction. [For full proceedings, see ED571332.]

This paper presents findings of a study that developed, measured, and tested the Organizational/Supervisory (O/S) Model of Instructional Supervision. The six dimensions of the Organizational Supervisory Climate Inventory (OSCI)--organizational structure, professional autonomy, collaborative sharing, district supervisory climate, reflection, and centralization--served as independent variables. The dependent variables were three recognized indices of effectiveness--school effectiveness, organizational effectiveness, and school holding power. Methodology involved a survey of 7,358 professional staff (primarily teachers) and 452 administrators in 1 state and interviews with 2 randomly selected teachers and the principal in 20 of the schools that participated in the survey. Findings indicate that the OSCI instrument is a highly reliable and stable measure of the organizational/supervisory climate, which indicates that the relationship between school effectiveness and school climate can be conceptualized and measured multidimensionally. The O/S model shows the usefulness of subscribing to a richer view of school instructional supervision as a fundamentally organizational phenomenon. Appendices include information on the research-design process, the O/S Model of Instructional Supervision, and statistical data. Contains 29 references. (LMI)

This paper presents findings of a study that sought to determine whether a composite score is a more appropriate dependent variable in the measurement of school effectiveness than any one component score. The viability of using a composite score, such as the School Effectiveness Index (SEI), was compared with the viability of using any one of five different subject areas. Specifically, the study examined the level of consistency that exists when school-effectiveness-classification models are held constant across test scores. Multiregression analysis of data from 315 Louisiana high schools was used to predict student achievement on a statewide-administered criterion-referenced test (CRT) from indices of socioeconomic status (SES), racial makeup, school report card variables, and demographic data. Findings indicate that a composite index incorporates the varied information that each component issue provides and presents a better overall picture of school effectiveness than would any one test component. In conclusion, a school's success in one subject does not necessarily ensure success in other areas. The composite score is preferable for school-effectiveness research, because it provides more information and is more reliable than a subtest. Two tables and one figure are included. (LMI)

This paper explores the PISA [Programme for International Student Assessment] achievement of twenty countries in light of some of their demographic differences. SES [student socioeconomic status], nuclear family, gender, home language, and native status were predictive of achievement for every country. Demographics accounted for as little as 8 percent to as much as 22 percent of individual score variance depending on the country and subject. Being male was almost a universal advantage in math, but was a far greater disadvantage in reading for every country. The relative performance of some countries changed when scores were adjusted for demographic differences; however, the Asian countries and Finland remained on top. Instructional strategies related to countries performing above expectations were explored.

The past few years has seen the rapid growth of data mining approaches for the analysis of data obtained from Massive Open Online Courses (MOOCs). The objectives of this study are to develop approaches to predict the scores a student may achieve on a given grade-related assessment based on information, considered as prior performance or prior activity in the course. We develop a personalized linear multiple regression (PLMR) model to predict the grade for a student, prior to attempting the assessment activity. The developed model is real-time and tracks the participation of a student within a MOOC (via click-stream server logs) and predicts the performance of a student on the next assessment within the course offering. We perform a comprehensive set of experiments on data obtained from two openEdX MOOCs via a Stanford University initiative. Our experimental results show the promise of the proposed approach in comparison to baseline approaches and also helps in identification of key features that are associated with the study habits and learning behaviors of students.

This research connects several data-driven educational data mining approaches to a framework for interaction developed in educational research. In particular, 10 million usage data points collected by a Learning Management System used by students and teachers in 450 online undergraduate courses were analyzed with this framework. A range of educational data mining techniques were employed, including K-means clustering, multiple regression, and classification, to both explore and predict student final grades and course completion rates. Findings show that support for the overall model varied with the way data were mapped to the framework (e.g., static vs. temporal features) and the analysis technique used (with clustering and classification providing more useful insights). [For the full proceedings, see ED592609.]

The goal of the current study was to determine whether Concept of Word in Text (COW-T) predicts later sight word reading achievement in Spanish, as it does in English. COW-T requires that children have beginning sound awareness, automatic recognition of letters and letter sounds, and the ability to coordinate these skills to finger point accurately to words in memorized text. Participants in the current study (n = 90) were students in bilingual or dual-language schools in Minnesota, Missouri, Virginia, and Washington D.C. who were receiving literacy instruction in Spanish. Students were administered six early literacy tasks (i.e., alphabet and digraph recognition, letter sound knowledge, beginning sound awareness, rhyme awareness, spelling, and COW-T) in spring of kindergarten, followed by measures of automatic sight word reading in fall and spring of 1st grade. Multiple regression analyses revealed that of the six early literacy tasks administered in kindergarten, COW-T had the highest correlation with both fall and spring 1st grade sight word reading, even when controlling for fall sight word reading in predicting spring sight word reading.

The purpose of this study was to investigate if students are prepared to release any personal data in order to inform learning analytics systems. Besides the well-documented benefits of learning analytics, serious concerns and challenges are associated with the application of these data driven systems. Most notably, empirical evidence regarding privacy issues for learning analytics is next to nothing. A total of 330 university students participated in an exploratory study confronting them with learning analytics systems and associated issues of control of data and sharing of information. Findings indicate that sharing of data for educational purposes is correlated to study related constructs, usage of Internet, awareness of control over data, and expected benefits from a learning analytics system. Based on the relationship between the willingness to release personal data for learning analytics systems and various constructs closely related to individual characteristics of students it is concluded that students need to be equally involved when implementing learning analytics systems at higher education institutions. [For the full proceedings, see ED562093.]