Previous events

At-risk youth, resilience, and academic outcomes

Professor Andrew J. Martin, School of Education, University of New South Wales

23 June 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall, Quantitative Methods Hub

One-day workshop on Models for diary data

Professor Bernhard Schmitz, University of Darmstadt & Dr. Lars-Erik Malmberg., Department of Education

19 June 2014 09:00 - 15:00
Manor Road, Oxford

Convener: Dr Lars Malmberg, Quantitative Hub
Contact: Thomas Day for further information and registration

Abstract
Diary studies have long been used in family research. The traditional method of diary studies is one daily report over an extended period of time. The advantages of diaries are (1) the closeness in time between an event and the report of the event, thus minimising retrospection bias, (2) the collection of multiple reports over an extended period of time, (3) the possibility of the diary to function as a self-monitoring tool, a key element of self-regulation, and (4) the possibilities to use these in intervention designs. Paper and pencil versions and electronic versions of diaries appear to work equally well with participants. Importantly, diary data allows us to model processes within individuals over time. Intervention designs can be incorporated using interrupted time-series models. Such models have been used in intervention studies with primary school children’s understanding of mathematics and university students’ self-regulation of learning. Early models of multivariate time-series of individual cases can be incorporated into multiple-group models.

In this one-day workshop we will cover state-of-the-art models for diary data, including descriptive analysis, multilevel designs, univariate and multivariate time-series designs. We will provide overviews of principles of research designs (e.g., research questions and hypotheses for diary data, and examples of diaries), data integrity and interpretation (data quality control, problems with aggregation, descriptives) and inferential methods (e.g., ARIMA, intervention analysis, dynamic relationships).

The instructors
Bernhard Schmitz,  Prof. Dr., holds a full professorship at the  department of educational psychology of the University of Technology Darmstadt.

He earned his doctoral degree at the Free University of Berlin, Germany and his habilitation at the University of Technology Berlin, Germany. He worked at the Max-Planck-Institute for Human Development in Berlin. His research deals with self-regulation, diaries and time-series analyses. He published numerous articles and books about time-series and self-regulation. He is member of the editorial board of Metacognition and Learning and Learning and Instruction.

Docent, Dr. Lars-Erik Malmberg, Associate Professor, Department of Education, University of Oxford, UK.

Lars-Erik Malmberg, Dr, is University Lecture in Quantitative Methods in Education at the Department of Education, University of Oxford, and Docent in Quantitative Methods at the Faculty of Education, Åbo Akademi University in Finland, and is associate editor of the Journal of Learning and Instruction. He investigates substantive educational research questions using state-of-the-art quantitative modelling, including multilevel and structural equation models. His recent research is on intraperson (situation-specific) aspects of students’ learning and teachers’ mastery experiences, stemming from two projects: the Learning Every Lesson (LEL) and Teaching Every Lesson (TEL) studies.

Worked examples
Bernhard Schmitz’s worked example deals with a sample of 361 students in grade 5 who worked with diaries for a period of 6 weeks. The students were trained with respect to self-regulated learning. Analyses of trends, intervention analyses and multivariate time -series will be demonstrated.

Lars Malmberg’s worked example of a diary study of teacher education students during practicum, and the Learning Every Lesson (LEL) study consisting 300+ students who completed the Learning Experience Questionnaire between 2 to 34 times during a week of school, including self-reports of e.g., effort exertion and competence beliefs. We will demonstrate e.g., the multilevel model for change, lagged analysis, and variations of the latent growth model.

Prerequisites
The course assumes some familiarity with applied statistical analysis of data up to the level of multiple regression, as would be commonly taught in most introductory UK graduate student courses. Some initial familiarity with R, SPSS, MPLUS is beneficial but not essential.

Schedule
The course will be held in the IT room of the Manor Road Building from 9am-3pm on Thursday 19th June 2014. We will take a lunch break at noon. Tea/coffee and sandwich lunches will be provided.

Sponsor
The seminar is organized by the Quantitative Methods Hub (Quant Hub), at the Department of Education, University of Oxford, and sponsored by the Social Science Division Teaching Excellency Award.

Registration
The course is free of charge, but we ask you to submit a brief statement of how the methodological workshop would fit your ongoing research. A certain number of seats are available so please book on time. The deadline for registration is Thursday 5th June. For enquiries and registration please contact Thomas Day, email: thomas.day@education.ox.ac.uk. Once we have accepted your brief statement we will confirm your registration. Once registered, please let us know if you will not be able to attend, as we can then reallocate your seat to someone else. Please note, early registration is advised as we are offering places on a first come first served basis, providing applicants meet the criteria.

Accommodation
Participants need to make their own accommodation and travel arrangements. You can find information about accommodation in Oxford on the Oxford City website or on the Oxford Daily Information website.

In the event of excess demand for places at this event the selection of participants will be based on a number of criteria. The following groups will be given priority:

- those engaged particularly in intensive longitudinal research
- those engaged in social science research

Download a registration form
If you have any questions about any aspect of the course please email thomas.day@education.ox.ac.uk

Trajectories of parental depressive symptoms and their impact on early child behavior problems

Dr. Martina Narayanan, Norwegian Center for Child Behavioral Development & Institute of Education, University of London

16 June 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall, Quantitative Methods Hub

One-day workshop on: Models for intensive longitudinal data

Theodore A. Walls, University of Rhode Island & Lars-Erik Malmberg, Department of Education

13 June 2014 09:00 - 15:00
Manor Road, Oxford

Convener: Dr Lars-Erik Malmberg, Quantitative Hub
Contact: Thomas Day for further information and registration

Abstract
Modern technology allows us to collect intensive longitudinal data, i.e., repeated records over extended periods of time, using techniques such as experience sampling, ecological momentary assessment, contextual activity sampling systems, and ambulatory biopsysiological recorders. These studies abound but few researchers have the design and data analytic skills needed to optimize opportunities from them. For example, when modelling intensive longitudinal data particular care needs to be taken with regard to the time-dimension of the data-collection (e.g., seconds., minutes, hours, days) and whether the data-points are random, fixed or event-driven. It is possible to integrate different time-scales of measurement within the same model by operationalising lower time-scale processes (e.g., minute to minute) and higher order time-scale processes (e.g., day to day) (Walls et al., 2011). Such techniques would allow the integration of data collection and analysis of both subjective reports and objective biophysiological measures. Researchers entering into formal academic posts need a strong command of both design and statistical fundamentals bearing on analysis of these large and complex databases. This workshop is intended to provide a beginner to intermediate level exposure toward advanced training in these areas.

In this one-day workshop we will cover state-of-the-art models for intensive longitudinal data, including descriptive analysis, multilevel designs and time-series designs. We will provide overviews of principles of research designs (e.g., research questions and hypotheses for intensive longitudinal data, data-collection and data-collection tools), data integrity (programming, data quality control, privacy, prevention of non-response) and inferential methods (e.g., descriptives, ideographic vs. nomothetic, contextual and multilevel, intervention elements, the role of time in the model).

The instructors
Dr. Theodore A Walls, Associate Professor, Department of Psychology, University of Rhode Island, CT, USA
Dr. Walls' research program is in applied quantitative methods, specifically in the development of methods for longitudinal data analysis for health behavioral and developmental studies. His recent work is focused on developing applied statistical methods, research designs and innovative empirical demonstrations in areas of health behavior research. In this work, he has utilized and extended approaches in the structural equation modeling, time series and multilevel modeling domains. This involves identifying, consolidating, developing, and disseminating statistical methods that can be used to analyze these data, particularly with respect to uncovering underlying functional processes. He is co-editor of a recent volume on longitudinal methods from Oxford University Press and has served extensively on advisory boards and panels regarding longitudinal design and analysis at NIH, NSF, NSERC and in the EU, Asia, and Russia. In 2006-2010, he was a featured speaker at institutes sponsored by NHLBI, the Psychometric Society, and the American Psychological Association, the Psychometric Society, and the American Psychological Society. He is a 2004 recipient of a Research Scholar Award from the Murray Research Center at Radcliffe. He is a 2007 recipient of a five year research scholar award from the American Cancer Society to develop engineering-based models of self-regulation of smoking behavior. His current work involves development of technologies and inferential methods for tracking of health behaviors ranging from smoking to surgical handwashing to oral health care.

Docent, Dr. Lars-Erik Malmberg, Associate Professor, Department of Education, University of Oxford, UK
Lars-Erik Malmberg, Dr, is Associate Professor in Quantitative Methods in Education at the Department of Education, University of Oxford, and Docent in Quantitative Methods at the Faculty of Education, Åbo Akademi University in Finland. He is associate editor of the Journal of Learning and Instruction. He investigates substantive educational research questions using state-of-the-art quantitative modelling, including multilevel and structural equation models. His recent research is on intraperson (situation-specific) aspects of students’ learning and teachers’ mastery experiences, stemming from two projects: the Learning Every Lesson (LEL) and Teaching Every Lesson (TEL) studies.

Worked examples
Dr. Walls will utilize, for example, data and/or designs reflecting affect during exercise, collected over three epochs and consisting of eight weeks of intensive measurement using sensors, electronic diaries and lab-based measurements. Typical constructs considered in this area include hedonic valence, arousal, social cognition and intention to exercise. In addition, other databases reflecting smoking behavior, alcohol use and other databases which include both physiological and psychological measures will be considered.

Dr. Malmberg’s worked example stems from the Learning Every Lesson (LEL) study consisting 300+ students who completed the Learning Experience Questionnaire between 2 to 34 times during a week of school, including self-reports of e.g., effort exertion and competence beliefs. The worked example covers variance components, intraperon correlations, multilevel confirmatory factor analysis, fixed and random effects models, and dynamic factor analysis.

Prerequisites
The course assumes some familiarity with applied statistical analysis of data up to the level of multiple regression, as would be commonly taught in most introductory UK graduate student courses. Some initial familiarity with R, SPSS, MPLUS is beneficial but not essential.

Schedule
The course will be held in the IT room of the Manor Road Building from 9am-3pm on Friday 13th June 2014. We will take a lunch break at noon. Tea/coffee and sandwich lunches will be provided.

Sponsor
The seminar is organized by the Quantitative Methods Hub (Quant Hub), at the Department of Education, University of Oxford, and sponsored by the Social Science Division Teaching Excellency Award.

Registration
The course is free of charge, but we ask you to submit a brief statement of how the methodological workshop would fit your ongoing research. A certain number of seats are available so please book on time. The deadline for registration is Thursday 5th June. For enquiries and registration please contact Thomas Day,

Once we have accepted your brief statement we will confirm your registration. Once registered, please let us know if you will not be able to attend, as we can then reallocate your seat to someone else. Please note, early registration is advised as we are offering places on a first come first served basis, providing applicants meet the criteria.

Accommodation
Participants need to make their own accommodation and travel arrangements. You can find information about accommodation in Oxford on the Oxford City website or on the Oxford Daily Information website.
In the event of excess demand for places at this event the selection of participants will be based on a number of criteria. The following groups will be given priority:
• those engaged particularly in intensive longitudinal research
• those engaged in social science research

Download a registration form
If you have any questions about any aspect of the course please email thomas.day@education.ox.ac.uk

Factor structure and measurement invariance across parental reports on the Strengths and Difficulties Questionnaire

Dr. Carlo Chiorri, Department of Educational Sciences, University of Genova

09 June 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall, Quantitative Methods Hub

The experience of traumatic events disrupts the stability of a posttraumatic stress scale

Dr. Miriam Lommen, Oxford Centre for Anxiety Disorders and Trauma (OxCADAT), Department of Experimental Psychology, University of Oxford

02 June 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall, Quantitative Methods Hub

Mapping tests: comparing upper secondary students' reading skills in L1 and L2

Lisbeth Brevik, Department of Teacher Education and School Research, University of Oslo

19 May 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall, Quantitative Methods Hub

Does an aptitude test affect socioeconomic and gender gaps in attendance at an elite university?

Jake Anders, Institute of Education, University of London

12 May 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall, Quantitative Methods Hub

Is the distribution of earnings in the UK shaped by educational attainment and occupational outcomes? An application of unconditional quantile regression techniques

Dr. Craig Holmes, Pembroke College

28 April 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall, Quantitative Methods Hub

Using smartphones to research daily life

Neal Lathia, University of Cambridge

10 March 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Steve Strand and Dr James Hall

Abstract
As smartphones proliferate throughout society, so too does the opportunity to use these devices to study, understand, and positively affect human behaviour in a variety of different contexts. Smartphone-based studies allows researchers to interact with their participants, via prompts to complete questionnaires, with less obtrusiveness than previous methods; moreover, these phones allow researchers to collect data from sensors on the phone that quantitatively encode behaviour. In this talk, I will review some of our recent work that uses a smartphone app to study daily moods: I will discuss the challenges of designing and deploying an app that has, to date, been downloaded approximately 30,000 times, and describe how behaviour can be quantified via sensor data. I will close by describing a new app that is being developed to apply our experiences to other domains beyond mood.

Challenges in developing teacher selection tools

Rob Klassen, University of York

03 March 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Steve Strand and Dr James Hall

Abstract
Research and theory in education and psychology provide some guidance about what makes for effective teaching, but developing reliable and valid teacher selection tools based on this body of knowledge presents a real challenge. In this talk I consider the challenges in developing teacher selection tools in the UK and internationally, and propose ways to improve selection practice.

An introduction to discrete Bayesian methods in educational research

Petri Nokelainen, University of Tampere, Finland

24 February 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Steve Strand and Dr James Hall

Abstract
The first part of the lecture gives an introduction to Bayesian statistics, which is interested in the probability of certainty that a given fact or proposition is true. The Bayesian way of calculating a probability is often labeled as ”subjective probability” or “inverse probability”, as its probability values ranging from zero (proposition is false) to one (proposition is true) are dependent on how much weight we are willing to lay on both the evidence and prior information available. Bayesian inference uses conditional probabilities to represent uncertainty. Conditional probability refers to a probability that one event will occur given that another has occurred.

The second part of the lecture discusses more specifically two Bayesian modeling techniques that allow analysis of small samples with nominal indicators and non-linear dependencies. The first technique, Bayesian Classification Modeling (BCM), allows a generic algorithm based selection of the best predictor variables for one class variable at the time. The second technique is called Bayesian Dependency Modeling (BDM), allowing construction of Bayesian Networks (BN). Practical applications of these techniques are discussed during the presentation.

Tracing students’ evolving activities and contextualized affects with the Contextual Activity Sampling System (CASS)

Hanni Muukkonen, University of Helsinki

17 February 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Steve Strand and Dr James Hall

Abstract
A longstanding question regarding learning in higher education has been how to examine patterns of evolving activities and how to account for the interplay of epistemic and emotional processes in their natural context. The Contextual Activity Sampling System (CASS) research methodology and the CASS-Query mobile application have been developed for contextually tracking of participants’ activities. The method relies on Ecological Momentary Assessment designed to trace the real-time advancement of learning activities by frequent sampling during periods of intensive follow-up. A study is presented in which 75 students from 3 universities took part in a two-week follow-up using mobile phones, with 5 queries per days, resulting in c. 3000 responses. Students replied what they were doing and rated the challenge, competence, commitment, absorption, interest, and the affects of irritation, anxiety, being energetic, and determination. A detailed qualitative content analysis was carried out to categorize activities during learning, working, and leisure. A contextualized examination of learning activities provided evidence on the particular patterns of meaningfulness and affects related to studying on one’s own, attending teaching or studying in collaboration. Methodological aspects on mobile data collection practices, the mixed methods approach, the analysis of datasets on evolving activities, and visualization of practices for the participants are discussed.

The hierarchical structure of work-related maladaptive personality traits

Nigel Guenole, Goldsmiths, University of London

03 February 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Steve Strand and Dr James Hall

Abstract
Important changes in how personality is conceptualized and measured are occurring in clinical psychology.  This talk focuses on one aspect of this work that work psychologists have been slow to embrace, namely, a new trait model that can be viewed as a maladaptive counterpart to the big five. I summarize the construction of a brief pathological personality measure, the G-50, designed to assist in the study of these substantive developments from clinical psychology in occupational settings. Responses to item pools assessing DSM-5 domain traits were collected from 696 working adults in England, Ireland, Wales and Scotland. Gender differences on domain traits were observed following invariance analyses while five-factor indicators projected into the latent space defined by pathological indicators revealed each big-five construct related to multiple pathological traits. Latent profile analyses revealed two classes, where a maladaptive class experienced worse outcomes on a range of job performance and health indicators. Support for a hierarchical factor structure was observed where DSM-5 domain traits are lower order indicators of internalizing and externalizing factors. Mixed evidence for a generalized psychopathology factor residing at the apex of the hierarchy was observed. Because lower-level maladaptive traits are described in the organizational sciences as 'Dark', we describe this generalized psychopathology factor as 'Black'.

Longitudinal effects of risk on children's problem behaviour from age 3 to 7: findings from the Millennium Cohort Study

Eirini Flouri, University of London

27 January 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Steve Strand and Dr James Hall

Abstract
Ecological and transactional theories link child outcomes to neighbourhood disadvantage, family poverty, and adverse life events. Traditionally, these three domains of risk have been examined independently of one another or combined into one cumulative risk index. The first approach results in poor prediction of child outcomes, and the second is not well rooted in ecological theory as it does not consider that distal risks (such as poverty) may indirectly impact children through proximal risks (such as adverse life events). In this study, we modelled simultaneously the longitudinal effects of cumulative risk in these three specific domains on children’s internalising and externalising problems, exploring the role of parenting in moderating these effects. Our sample followed 16,916 children (at ages 3, 5, and 7 years) from the UK Millennium Cohort Study. Parenting was characterised by parent-child relationship, involvement in learning, and negative discipline. We found that neighbourhood disadvantage, family poverty and adverse events were all simultaneously related to the trajectories of both outcomes. As expected, parenting moderated, not mediated, risk effects. A positive parent-child relationship, rather than greater involvement or authoritative discipline, most consistently ‘buffered’ risk effects. A good parent-child relationship appears to promote young children’s emotional and behavioural resilience to different types of environmental risk.

Conceptualising interaction and learning in Massive Open Online Courses (MOOCs)

Nabeel Gillani, Department of Education

20 January 2014 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Steve Strand and Dr James Hall

Abstract
Massive Open Online Courses (MOOCs) enable lifelong learners from around the world to interact with one another at unprecedented scales. Early literature on MOOCs has investigated the nature of learner interactions with their course environments. However, to date we know very little about the nature of interactions between learners or how these individuals exchange information with one another. Through a mixed method analysis of a MOOC that emphasizes collaborative problem-solving efforts, we aim to better understand who interacts with who in MOOCs, and how. We plan to interpret these interactions by contextualizing them according to the demographic characteristics and academic activities of each learner. These investigations will aid in analysing the formation of crowds versus communities in discussion settings; how information is aggregated and transmitted through interaction networks; and how participant backgrounds, course activities, performance, and communication tendencies are related. Using social network analysis in conjunction with insights derived from observations, participant interviews, and surveys, we hope to uncover how interaction patterns help us to understand how learning occurs through online interactions in ways that build on existing theoretical frameworks developed from previous learning and technology research. Ultimately, we aim to use this hybrid analytical framework to develop a typology that reflects the different ways in which MOOC participants communicate and interact in order to learn.

The mediating role of early literacy skills in the relationship between family SES and academic achievement: an investigation of causal mechanisms with international assessment data

Dr. Daniel Caro & Lorena Orega Ferrand, Department of Education, University of Oxford

02 December 2013 12:15 - 13:45
Seminar Room J

Convened by Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall

Following two Quant SIG presentations on causality in educational studies with observational data, this third presentation provides an applied example on the study of causal mechanisms. The example with international assessment data looks at the mediating role of early literacy skills in the relationship between family SES and academic achievement. The international assessment data has a cross-sectional design, but questions on early literacy skills refer to the period before entering school and family SES precedes both early literacy skills and academic achievement. From this perspective, the retrospective data provides an interesting opportunity to address a causal question cross-sectionally. The methodology integrates traditional mediation analysis with the potential outcomes framework. With that, it allow us to interpret mediation results causally.

Does homework make a difference to achievement?

Professor Pam Sammons, Department of Education, University of Oxford

25 November 2013 12:15 - 13:45
Seminar Room J

Convened by Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall

Abstract
The relationships between the time students report they spend on homework on a typical week night in Year 9, their self-regulation and academic attainment and progress in Key Stage 3 (age 14) of English secondary education are explored. Although the links between homework and academic outcomes have been the subject of research in many countries and different phases of education, the conclusions are not always convergent. A meta-analysis of the U.S. research showed a modest positive effect of homework on academic outcomes, stronger in the middle and high school (Cooper, Robinson, & Patall, 2006). The present paper used MLM and SEM to investigate the relationships between time students say they spend on homework and academic achievement and progress in English, mathematics and science during secondary school using data from the longitudinal educational effectiveness study (EPPSE) conducted in England. Multilevel analyses showed that time spent on homework is a statistically significant, positive and moderately strong predictor of attainment in all core subjects in KS3 after control for the influence of students’ individual (age, gender, birth weight etc.), family (SES, FSM, salary) and home learning characteristics and school context. Similar effects are found for analyses of student progress across KS3. These also controlled for student prior attainment in Year 6 in the models. Additionally, strong effects for time spent on homework were obtained in further models that included measures of the students’ perceptions of their secondary school’s emphasis on learning and behavioural climate. SEM modeling explored possible causal relationships between time spent on homework and academic outcomes in year 9 and the role of prior self-regulation (year 6). The models identify direct and indirect relationships between gender, mother’s qualification level, self-regulation, time spent on homework and academic outcomes.

Selective education in England

Professor David Jesson, University of York

18 November 2013 12:15 - 13:45
Seminar Room J

Convened by Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall

Abstract
The 164 grammar schools represent a small minority of the 3000+ secondary schools which provide education (mainly) from age 11 to 16 and beyond. These schools admit around 22,000 pupils annually – some 4% of the total cohort on state secondary schools. The fact that whilst less than 3% of pupils in receipt of Free School Meals currently gain access to Grammar schools over 13% of those admitted come from outside the state-sector (mainly from private ‘Prep’ schools?) suggests a degree of imbalance in the operation of current entry policies. This factor varies substantially between schools and whilst not a completely new finding, has focused an unprecedented degree of attention on the nature of Grammar schools admissions and on wider questions of social mobility. The importance of these schools socially and politically, however, is out of all proportion to their numbers and they continue to exercise a ‘fascination’ for commentators from both the political left and right of the political spectrum. Their educational importance is also a highly contested area and has been the focus of substantial, and it has to be said, inconclusive conclusions as to their impact on pupils’ performance. The Sutton Trust has recently commissioned further research on the issue of the intakes of pupils to Grammar schools. This was published last week and Professor Jesson’s session will review some of the historical, social and quantitative evidence in this Report related to admissions to these schools. Grammar schools and the local authorities in which they are placed are categorised and their composition and comparison with other schools in these areas will be discussed, along with substantial changes which are beginning to happen within many current Grammar schools

The specific aspect of ‘disadvantaged’ pupils’ gaining access to Grammar schools will form a major feature of the analyses.

Uncertainty in the measurement of school effects and peer effects in the English secondary school system

Dr. John Fletcher, Department of Education, University of Oxford

11 November 2013 12:15 - 13:45
Seminar Room J

Convened by Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall

Abstract
Comparisons of institutional performance using value-added models can be deficient in terms of the underpinning hierarchical model or the data on which the prediction of performance is based. An overview of the weaknesses is essential to decisions about improving any model of this kind. The Contextual Value-Added model for English secondary schools (ECVA model) provides a case study for assessing the relative importance of different problems in contextual models. I conclude that the overriding problem in the ECVA model is a lack of data to support the differentiation of variation in teacher performance and variation in school performance.

Student evaluations of university teaching: recommendations for policy and practice (Public Seminar)

Professor Herb Marsh, Department of Education and University of Western Sydney

04 November 2013 17:00 - 18:30
Seminar Room A

Convener: Professor Steve Strand, Quantitative Methods Special Interest Group

Abstract
Students' evaluations of teaching effectiveness (SETs) have been the topic of considerable interest and a great deal of research in universities all over the world. Based on reviews of research by myself and others, SETs are: multidimensional; reliable and stable; primarily a function of the instructor who teaches a course rather than the course that is taught; relatively valid against a variety of indicators of effective teaching; relatively unaffected by a variety of variables hypothesized as potential biases, such as grading leniency, class size, workload and prior subject interest;  and demonstrably useful in improving teaching effectiveness when coupled with appropriate consultation.

Although SETs have a solid research base stemming largely from research conducted in the 1980s, it is surprising that research conducted in the last decade has not done more to address critical limitations previously identified and incorporate exciting methodological advances that are relevant to SET research. Perhaps the most damning observation is that most of the emphasis on the use of SETs is for personnel decisions rather than on improving teaching effectiveness. Why do universities continue to collect and disseminate potentially demoralising feedback to academics without more fully implementing programs to improve teaching effectiveness? Why is there not more SET research on how to enhance the usefulness of SETs as part of a program to improve university teaching? Why have there been so few intervention studies in the last decade that address the problems identified in reviews of this research conducted a decade ago? These, and other issues, are addressed in this presentation.

Professor Herb Marsh holds a joint appointment at the Centre for Positive Psychology and Education at the University of Western Sydney and at Oxford University. He is an “ISI highly cited researcher” (http://isihighlycited.com/) with 340 publications listed in the World of Science with more than 18,000 citations, and an ISI-H-index = 69, and recently achieved a Google Scholar H-Index of 100. He founded and Directs the SELF Research Centre that has 450 members and satellite centres at leading Universities around the world, and co-edits the SELF monograph series. He coined the phrase substantive-methodological research synergy which underpins his research efforts.  His major Research/Scholarly interests include self-concept and motivational constructs; evaluations of teaching effectiveness; developmental psychology, quantitative analysis; value-added and contextual models; sports psychology; the peer review process; gender differences; peer support and anti-bullying.

Do schools reflect, compensate for, or exacerbate inequalities in society?

Professor Stephen Gorard, University of Durham

04 November 2013 12:15 - 13:45
Seminar Room J

Convened by Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall

Abstract
This paper continues an on-going investigation of the social and economic ‘segregation’ of students between schools in England, and of the likely causes of the levels of and changes over time in that segregation. The data presented here come from a re-analysis of the intakes to all mainstream schools in England 1989-2012 as portrayed by the official returns to the Annual Schools Census. Using a segregation index it shows how strongly clustered the students are in particular schools in terms of six indicators of potential disadvantage – representing poverty, learning difficulties, first language and ethnicity. The results are presented for England, the Economic regions, and for local education authority areas. The paper shows again, and with further years than previously, that each indicator has its own level and pattern of change over time. This suggests that there is not just one process of segregation. However, the patterns for primary-age schools (5-10) are exactly the same for most indicators as the patterns for secondary-age schools (11-18). These two findings in combination effectively rule out a large number of potential explanations either for changes in or levels of segregation - including volatility of small numbers, and recent changes in the types of schools and in the ways in which school places are allocated. Instead, based on correlations with other indicators of population, school numbers, and the economy, a new set of determinants is proposed. The long-term underlying level of segregation appears to be the outcome of structural and geographic factors. However, the annual changes in segregation for most indicators can be explained most simply by changes in the prevalence of each indicator. For example, the UK policy of inclusion has considerably increased the number of students with statements of special needs in mainstream schools, and this has resulted, intentionally, in less segregation in terms of this indicator. Segregation by poverty, however, requires something further to explain changes over time, and this is provided at least partly by changes in GDP over time, and partly as a one-off impact of increased parental choice. Some of these factors, such as the global economy or the prevalence of specific ethnic minority groups, are not directly under policy-makers’ control. This means that it is the more malleable factors leading to the underlying levels of poverty segregation that should be addressed by any state wanting a fair and mixed national school system. In England, these controllable factors include the use of proximity to decide contested places at schools, the growth of Academies, and the continued existence of faith-based and selective schools. The prevalence of Academies in any area is strongly associated with local levels of SES segregation, and this is especially true of the more recent Converter Academies.

Application of several latent trait models (IRT) to the detection of rater effects.

Dr. Ed Wolfe, Principal Research Scientist, Pearson Education

28 October 2013 12:15 - 13:45
Seminar Room J

Convened by Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall

To date, much of the psychometric research concerning rater effects has focused on rater severity/leniency. Consequently, other potentially important rater effects have largely ignored by those conducting operational scoring projects. In this presentation, I will summarize a line of research that seeks to determine how and how well latent trait measurement models (AKA item response models) can be used to detect rater centrality and rater inaccuracy. Specifically, I will summarize the results of two data simulation studies designed to evaluate the Type I and Type II error rates of several statistical indicators that are implemented in several item response theory models. I will also describe the context within which analyses such as these can be used to improve the quality of human ratings in educational testing.

Exploratory structural equation modelling: an integration of the best features of exploratory and confirmatory factor analysis

Professor Herb Marsh, University of Western Sydney

21 October 2013 12:15 - 13:45
Seminar Room J

Convened by Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall

Exploratory and confirmatory factor analysis (EFA and CFA), path analysis, and structural equation modelling (SEM) have long histories in social science research. Although CFA has largely superseded EFA, CFAs of multidimensional constructs typically fail to meet standards of good measurement: goodness of fit, measurement invariance, lack of differential item functioning, and well differentiated factors in support of discriminant validity. Part of the problem is undue reliance on overly restrictive CFAs in which each item loads on only one factor. ESEM, an overarching integration of the best aspects of CFA/SEM and traditional EFA, provides confirmatory tests of a priori factor structures, relations between latent factors, multigroup/multi-occasion tests of full (mean structure) measurement invariance, incorporating all combinations of CFA factors, ESEM factors, covariates, grouping/MIMIC variables, latent growth, and complex structures that have typically required CFA/SEM. Due to misfit associated with overly restrictive measurement models with no cross-loadings, CFAs typically produce inflated factor correlations compared to ESEMs and to known population values for simulated data. This detracts from discriminant validity, undermines diagnostic usefulness, and results in complicated biases in more complex models. Hence, applied researchers are recommended routinely to conduct preliminary analyses at the level of individual items, comparing of ESEM and CFA measurement models based on all constructs to be considered in order to compare the suitability of CFA/SEMs and ESEMS for subsequent analyses.

Home and school contributions to the urban-rural cognitive achievement gap in Peru

Juan Castro, Department of International Development, University of Oxford

14 October 2013 11:15 - 12:45
Seminar Room J

Convened by Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall

This research seeks to measure the relative importance of home and school influences for the emergence of cognitive achievement gaps between urban and rural children in Peru. It proposes a general linear model for the production of cognitive skill and uses it to assess the sources of bias in different empirical specifications and to decompose the urban-rural gap in cognitive achievement distinguishing between contemporaneous and cumulative effects. The Young Lives study provides longitudinal information on children’s cognitive achievement and household characteristics, and results from a comprehensive school quality survey. Consistent estimation of all production function parameters is problematic due to the multiple sources of bias. Experimental or quasi-experimental methods are impractical due the diversity of inputs. Therefore, the proposed empirical strategy provides a ranking of effects and upper and lower bounds for the relative contributions of home and school inputs.

Academic buoyancy, psychological risk, and academic outcomes: quantitative approaches to reciprocal relationships and processes over time

Professor Andrew J. Martin, Faculty of Education and Social Work, University of Sydney

07 October 2013 12:15 - 13:45
Seminar Room J

Convened by Professor Steve Strand, Dr Lars-Erik Malmberg and Dr James Hall

Abstract
Academic buoyancy refers to students’ capacity to effectively deal with ‘everyday’ academic setback in the ordinary course of school life (e.g., study pressure, a poor result, competing deadlines, difficult schoolwork). In this session, two studies are presented, both drawing on data from 2,971 students (11-19 years) from 21 high schools at two time waves across a one-year interval. Study 1 uses cross-lagged structural equation models to examine the relative salience of (1) prior academic buoyancy in predicting subsequent psychological risk (academic anxiety, failure avoidance, uncertain control, emotional instability, neuroticism) and (2) prior psychological risk in predicting subsequent academic buoyancy. Study 2 more closely explores the processes relevant to academic buoyancy as it relates to academic outcomes. Using structural equation modelling, it explores academic buoyancy in the context of literacy, numeracy, and students’ perceived control. The findings hold applied and conceptual implications for practitioners and researchers seeking to help students deal more effectively with adversity in school life.

Modelling the effects of contextual risk and temperament on child psychopathology and ability

Dr. Kim Drake, University of West London

10 June 2013 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons & Dr James Hall, Quantitative Methods SIG

Evidence consistently appears to implicate contextual risk (CR) as a contributing factor in child functional outcomes.  A problem though with existing work is that, to measure CR, researchers have often included measures of socio-economic environment, which research has shown to be problematic when used interchangeably as indices of difference/disadvantage.  Cumulative CR measures, enabling the consideration of both individuals’ socio-economic environment as well as their relative position in society (i.e. socio-economic disadvantage; SED), seem more reliable.  Relatively little is known about the role of SED in child psychopathology and/or ability. Moreover, children with problematic temperaments tend to be over-represented within socio-economically disadvantaged families.  Using a multivariate structural equation modelling approach on the Millennium Cohort Data, this paper examines the role of child temperament as a moderator in the relationship between the effects of SED on psychopathology and ability.  The results and implications there-from will be discussed.

Social network analyses in the educational science

Dr. Ralf Woelfe, Department of Experimental Psychology

03 June 2013 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons & Dr James Hall, Quantitative Methods SIG

Within the last decade, social scientists increasingly recognize the potential of social network analyses (SNA), which enrich the explanation of human behaviour by explicitly taking their social structure into account. Driven by the recent technical advancement of statistical programs that allow the application of complicated algorithms to large datasets, SNA have reached a point of analytic refinement that make them a valuable tool in order to test the social mechanisms that underlie our behaviour. Particularly for educational research, this analytic perspective is highly valuable, because most mechanisms in the educational setting are operating in contextual structures and the main unit of analysis (students) is especially susceptible to influence processes and social dynamics. Therefore, in this presentation, I aim to provide a short introduction of SNA by highlighting their potential and limitations tailored to the specific interests of educational scientists.

Language learner strategies in teacher-student interaction

Daniel Fung, Department of Education

20 May 2013 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons & Dr James Hall, Quantitative Methods SIG

How effective are educational systems? A value-added approach to study trends in international large scale assessments

Dr. Jenny Leinkeit, Department of Child Development and Education, University of Amsterdam

13 May 2013 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons & Dr James Hall, Quantitative Methods SIG

The paper investigates how effectively educational systems grow, i.e. change, in their performance by applying conceptual and methodological approaches known from educational effectiveness research. The modelling strategy is an adaption of Willms and Raudenbush (1989) who nested students into cohorts and cohorts into schools to examine the stability of school effects on levels of educational attainment. International large scale assessments have a similar multilevel design which nests students into schools, schools into cohorts (i.e. survey cycles), and cohorts into educational systems. The proposed model evaluates educational system effectiveness for different cohorts of students across systems and over time. Data from the Progress in International Reading Literacy Study (PIRLS) and the Programme for International Student Assessment (PISA) trend systems are analysed. Results change the picture of “high” and “low” performing systems, when contextual conditions and prior performances are taken into account.

Causal inference in educational studies with observational data

Dr. Daniel Caro, Oxford University Centre for Educational Assessment

06 May 2013 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons & Dr James Hall, Quantitative Methods SIG

Value-added modelling of teacher effects in Chile: researching magnitude, consistency and predictors

Lorena Ortega Ferrand, Department of Education

29 April 2013 12:15 - 13:45
Seminar Room J, 28 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons & Dr James Hall, Quantitative Methods SIG

Reading and language interventions: using randomised controlled trials to inform theory and practice

Dr. Fiona Duff, Department of Experimental Psychology

22 April 2013 12:15 - 13:45
Seminar Room J, 30 Norham Gardens

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons & Dr James Hall, Quantitative Methods SIG

Theoretically-motivated interventions that are evaluated using robust methodologies and robust statistical analyses have much to offer in terms of our understanding of theory and practice.  This talk will focus on intervention programmes that aim to improve the reading and language skills of two groups of children: children who enter school with language weaknesses, and children with Down syndrome.  Two studies will be presented, both of which combine a randomised controlled trial methodology with regression analyses.  The resultant implications for psychological theory and educational practice will be discussed.

Applications of propensity matching in measurement invariance studies

Dr Carlo Chiorri, University of Genoa

04 March 2013 12:15 - 13:45
Seminar Room J

This is a Quant Sig Seminar convened by Dr Lars-Erik Malmberg and co-convened by Professor Pam Sammons and Dr James Hall

Mediation, moderation, and interaction: definitions, discrimination & (some) means of testing

Dr James Hall, Department of Education

25 February 2013 12:15 - 13:45
Seminar Room J

This is a Quant Sig Seminar convened by Dr Lars-Erik Malmberg and co-convened by Professor Pam Sammons and Dr James Hall

As more complex statistical analyses become available in software packages, the terms “Mediation”, “Moderation” and “(Statistical) Interaction” see increased use.  However, uncertainty continues over their definitions which manifests as inconsistent guidelines and different means of testing.  Further, there has been a lack of guidance provided specifically for Educational Researchers.  To address these problems, this presentation provides clear definitions that discriminate key-terms, note real-life ambiguities particular to education and documents the various means of testing that are available to researchers.  The presentation ends with an exemplar Moderation from educational research that is tested with three alternative statistical procedures that then have their results compared and contrasted.

Time-series in practice: seasonal behaviour and cross-sectional analyses

Dr Orlaith Burke, Department of Statistics

18 February 2013 12:15 - 13:45
Seminar Room J

This is a Quant Sig Seminar convened by Dr Lars-Erik Malmberg and co-convened by Professor Pam Sammons and Dr James Hall

Time-series data provides plenty of scope for new and interesting forms of analysis. The statistical methodologies presented in this talk focus on the modelling of seasonal behaviour and the analysis of time-series cross-sectional data. These methods are motivated by an application in radiological protection.

The average indoor radon concentration in Ireland is among the highest in the world, causing approximately 150-200 lung-cancer deaths each year in Ireland alone. In order to control exposure, it is necessary to ensure accurate measurement policies are in place. Radon concentrations are affected by seasonal changes, regional variation and external factors such as climate and geology. Therefore, several different time-series methods are required to answer our questions about the behaviour of indoor radon concentrations.

Family socio-economic disadvantage and children’s emotional and behavioural development: the importance of self-control

Dr. Emily J. Midouhas, Institute of Education, University of London

11 February 2013 12:15 - 13:45
Seminar Room J

This is a Quant Sig Seminar convened by Dr Lars-Erik Malmberg and co-convened by Professor Pam Sammons and Dr James Hall

Family poverty is strongly related to children’s emotional and behavioural problems. Cognitive ability and self-control have been related to children’s emotional/behavioural resilience to family poverty, but have not been examined jointly as factors promoting resilience, despite their interrelatedness. In this study, we investigated the role of two aspects of self-control - self-regulation and low emotional dysregulation – along with verbal cognitive ability in children’s emotional/behavioural resilience to family poverty from early to middle childhood (ages 3, 5, 7; N = 16,916). Using multivariate response growth curve modelling, we found that the relationship between poverty and internalising and externalising problems was stronger for children with higher levels of emotional dysregulation, whereas poor children with high levels of self-regulation appeared indistinguishable from affluent children with high levels of self-regulation in the rate of growth of internalising and externalising problems. Verbal cognitive ability moderated the association between poverty and internalising problems only. Neither self-control nor verbal cognitive ability mediated the association between family poverty and children’s internalising or externalising problems. Self-control seems to be an important protective factor for children growing up in poor families.

Markov chain Monte Carlo methods for multiple interventions meta-analysis in psychosocial intervention

G.J. Melendez-Torres, University of Oxford Centre for Evidence-Based Intervention

04 February 2013 12:15 - 13:45
Seminar Room J

This is a Quant Sig Seminar convened by Dr Lars-Erik Malmberg and co-convened by Professor Pam Sammons and Dr James Hall

Multiple interventions meta-analysis is an emerging analytic tool that allows researchers, practitioners, and policymakers to move beyond the question 'does this intervention work?' to 'which intervention works best?'  Though multiple interventions meta-analysis has been called the 'next generation evidence synthesis tool' (Salanti 2012), its application has until very recently been limited to studies of pharmacological and other clinical interventions.  Indeed, one of its difficulties is that it is often implemented using Markov chain Monte Carlo analysis, a method that, though powerful and commonly used, is a challenge to understand and implement.  This presentation will introduce the benefits and challenges of this evidence synthesis methodology, review the Markov chain Monte Carlo principles underlying its implementation, and chart a line of inquiry for its adaptation to psychosocial interventions.

Do some schools narrow the gap? Differential school effectiveness by ethnicity, gender, poverty and prior attainment

Professor Steve Strand, Department of Education

28 January 2013 12:15 - 13:45
Seminar Room J

This is a Quant Sig Seminar convened by Dr Lars-Erik Malmberg and co-convened by Professor Pam Sammons and Dr James Hall

Abstract
This study analyses the educational progress of an entire national cohort of over 530,000 pupils in England between age 7 in 2000 and age 11 in 2004. The results show that Black Caribbean boys not entitled to free school meals, and particularly the more able pupils, made significantly less progress than their White British peers. There is no evidence that the gap results from Black Caribbean pupils attending less effective schools. There is also no evidence of differential effectiveness in relation to ethnic group; schools that were strong in facilitating the progress of White British pupils were equally strong in facilitating the progress of Black Caribbean pupils. There was some evidence of differential school effectiveness by pupil prior achievement, gender and poverty, but the absolute size of effects were small. The results suggests that the poor progress of Black Caribbean pupils reflects a systemic issue rather than the influence of a small number of ‘low quality’ schools.

Causal inference in educational studies with observational data

Dr. Daniel Caro, Oxford University Centre for Educational Assessment (OUCEA)

21 January 2013 12:15 - 13:45
Seminar Room J

This is a Quant Sig Seminar convened by Dr Lars-Erik Malmberg and co-convened by Professor Pam Sammons and Dr James Hall

The dynamics of flow experience: a person centered approach

Professor Katariina Salmela-Aro (University of Helsinki, Finland / Institute of Education, London

26 November 2012 12:15 - 13:45
Seminar Room J

A seminar in the Quantitative Methods SIG programme for the Michaelmas Term 2012

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons and Dr James Hall

Marking of national examinations: how reliable is it?

Professor Jo-Anne Baird, Oxford University Centre for Educational Assessment

19 November 2012 12:15 - 13:45
Seminar Room J

A seminar in the Quantitative Methods SIG programme for the Michaelmas Term 2012

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons and Dr James Hall

There are a number of sources of unreliability of national examinations:
-the tests and questions themselves (internal reliability)
-occasion-related factors (e.g. health, noisy environment)
-consistency of standard-setting (conversion of scores to grades) and markers (or raters).

This seminar will focus upon marker effects upon scoring and will introduce the literature on this topic before outlining studies on the scoring of public examinations in the UK.  Several studies have been conducted using multilevel modeling.  A study on the investigation of the qualifications needed to score reliably will be presented and, depending upon the time available, we might be able to discuss further studies on rater severity drift over time, effects of rater experience and the training of raters in groups.

School value-added and equity in Vietnam: the impact of peer performance on child school primary school attainment in Vietnam

Dr. Caine Rolleston, Young Lives project, University of Oxford

12 November 2012 12:15 - 13:45
Seminar Room J

A seminar in the Quantitative Methods SIG programme for the Michaelmas Term 2012

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons and Dr James Hall

Vietnam’s rapid economic and educational development is linked to a raft of liberalisation policies, which in the education sector include measures to ‘socialise’ the responsibility for education provision, including through increasing the financial contributions of households.  While growth has reduced poverty considerably, differences in educational quality and in households’ ability to pay vary widely between urban and rural locations and between regions of Vietnam.  Young Lives collected data in 2011at 80 schools in 5 provinces, focused on education in Grade 5.  This data is linked to detailed longitudinal data on the children’s development collected from their households since 2002.  This paper employs these data to examine the linkages between growing up in poverty and educational outcomes, exploring implications for social mobility in Vietnam.  It addresses the impact of poverty on access to quality schooling and the relationships between household resources and children’s achievement. Furthermore, Dr. Rolleston will also discuss peer effects in primary schools in Vietnam, in terms of its implications on academic attainment and child performance.

Using behavioural indicators of teacher engagement and emotions

Professor Rob Klassen, University of York

05 November 2012 12:15 - 13:45
Seminar Room J

A seminar in the Quantitative Methods SIG programme for the Michaelmas Term 2012

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons and Dr James Hall

The building blocks of collective decision-making under uncertainty

Dan Bang, University of Oxford, Department of Experimental Psychology and Calleva Research Centre

29 October 2012 12:15 - 13:45
Seminar Room J

A seminar in the Quantitative Methods SIG programme for the Michaelmas Term 2012

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons and Dr James Hall

In everyday life, many people believe that two heads are better than one. Indeed, our ability to solve problems together appears to be fundamental to the success and the future survival of the human species. But are two heads really better than one? I will address this question in the context of a novel experimental paradigm in which pairs of individuals make collective decisions about the occurrence of a faint visual target. I will report on computational and empirical findings to identify the building blocks of successful collective decision-making. In addition, I will report preliminary findings from a study in which we compared ‘children’ (aged 10-12) with ‘adolescents’ (aged 14-16) in order to explore the question: when do two heads become better than one? This talk requires no strong background in quantitative methods.

If you do not have access to the Department then please email and we will arrange access (QuantSIG@education.ox.ac.uk).

Having central A-level examinations in Germany: the effects on grading and on the comparability of grades

Dr. Monika Holmeier, University of Zürich

22 October 2012 12:15 - 13:45
Seminar Room J

A seminar in the Quantitative Methods SIG programme for the Michaelmas Term 2012

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons and Dr James Hall

Recently almost all German states have put into place centrally administered A-level examinations. Under this new system, the examination questions and grading criteria are no longer developed by the teachers themselves but by external examining bodies. As well as improving students’ levels of attainment these changes are intended to encourage the teacher to make use of grading processes based on the externally-set grading criteria and to improve the ways in which exam grades can be compared between students, classes and schools.

This presentation looks at the changes in the German A-level examination system and considers the extent to which teachers take into account the new compulsory grading criteria that they now have to follow. It will also look at students’ awareness of the criteria. Furthermore, the presentation shows whether students’ gender, ethnicity and socio-economic status have an effect on their examination grades (controlled for individual achievement), and how these effects might be altered in a central examination system.

Academic adversity and uncertainty examining students’ academic buoyancy, academic resilience, and adaptability.

Professor Andrew Martin, University of Sydney

15 October 2012 12:15 - 13:45
Seminar Room J

A seminar in the Quantitative Methods SIG programme for the Michaelmas Term 2012

Conveners: Dr Lars-Erik Malmberg, Professor Pam Sammons and Dr James Hall

Quant SIG Seminar

John Fletcher, Department of Education, University of Oxford

11 June 2012 12:15 - 14:00
Seminar Room J

A Quant Sig Seminar
Conveners: Professor Pam Sammons, Dr Lars-Erik Malmberg and Dr James Hall

Exploring students’ social-behavioural development and dispositions at secondary school : results from the longitudinal EPPSE research in KS3

Professor Pam Sammons & Professor Kathy Sylva, Department of Education, University of Oxford

28 May 2012 12:15 - 14:00
Seminar Room J

A Quant Sig Seminar
Conveners: Professor Pam Sammons, Dr Lars-Erik Malmberg and Dr James Hall

Educational inequality and outcomes of 8-15 year olds in Ethiopia

Nardos Tesfay, Department of Education, University of Oxford

21 May 2012 12:15 - 14:00
Seminar Room J

A Quant Sig Seminar
Conveners: Professor Pam Sammons, Dr Lars-Erik Malmberg and Dr James Hall

Patterns of Child Care Arrangements from Birth to 51 Months

Suna Eryigit (QuantSIG Seminar)

14 May 2012 12:15 - 14:00
Seminar Room J

A Quant Sig Seminar
Conveners: Professor Pam Sammons, Dr Lars-Erik Malmberg and Dr James Hall

Abstract: 
Conceptualizing childcare as a grouping variable based on its longitudinal patterns is an alternative to methods used in the childcare literature, such as existence, onset and amount of childcare. This study aims to investigate whether identifiable patterns of childcare arrangement exist from birth to 51 months and whether these patterns moderate the associations among cognitive development, maternal stimulation and difficult temperament. Six prevailing patterns of childcare arrangements with a cut-off at 36 months were identified based on initial analyses: continuous maternal (11%), continuous home (15%), maternal to centre (25%), home to centre (13%), continuous centre (15%) and multi type of care groups (22%). This talk will focus both on previous findings and new analyses conducted to investigate the longitudinal patterns of arrangements in each type of care using GMM.

Does teaching affect student’s chances of completing a degree at British Universities?

Andrea Canales, DPhil candidate, Department of Sociology

30 April 2012 12:15 - 14:00
Seminar Room J

Please contact Patrick Alexander for further information and/or access to the seminar room.

Abstract

This paper examines the effects of teaching (student-faculty ratio, learning resources and satisfaction with teaching quality) on student’s chances of completing a degree at British Universities.

I employ three data sources to conduct the empirical analyses . First, I use student-level longitudinal data provided by the Higher Education Statistic Agency (HESA), which contains information about student’s academic progression at British Universities throughout years. I also employ institutional-aggregated level data drawn from the National Student Survey (NSS), which provides information about student’s satisfaction with teaching quality at universities in the UK. Finally, I employ institutional information provided by HEFCE and HESA about universities’ learning resources (faculty and books) and infrastructure.

A series of factorial analyses were carried out with the questions from the National Student Survey (NSS), in order to determine how many dimensions composed student’s evaluation of teaching quality. Three dimensions were identified as central aspects of teaching quality: teaching organization, teaching feedback and learning resources. Once these dimensions were identified, multilevel binomial regression analyses (random intercepts) were used to determine the main effects of teaching and institutional variables on student’s attainment. Random coefficient models with cross level interactions were also conducted to analyze how these institutional aspects vary across different groups of students.

Overall, the analyses revealed that institutional aspects –those related to conventional characteristics and teaching aspects- contribute to explain some institutional variation in the student’s chances of degree completion. Likewise, the results confirmed that conventional features such as institutional selectivity or type of university have greater significance on student’s attainment than teaching aspects. The effect of teaching is rather small and varies across different student’s populations.

Quant Sig Seminar

Professor Herb Marsh (via Skype)

05 March 2012 12:15 - 14:00
Seminar Room J

Conveners: Professor Pam Sammons, Dr Lars-Erik Malmberg and Dr James Hall

Integrating Structural Equation models with Contextual Value Added models and Regression Discontinuity Designs: A progress report on two distinct studies.

Ioulia Televantou, Department of Education

27 February 2012 12:15 - 14:00
Seminar Room J

Conveners: Professor Pam Sammons, Dr Lars-Erik Malmberg and Dr James Hall

If you do not have access to the building, please contact Patrick Alexander to arrange access.

Protecting the development of 5-11 year olds from the impacts of early disadvantage: The Role of Primary School Academic Effectiveness

Professor Pam Sammons, Department of Education

20 February 2012 12:15 - 14:00
Seminar Room J

Conveners: Professor Pam Sammons, Dr Lars-Erik Malmberg and Dr James Hall

Abstract
Whether or not more effective schools can successfully mitigate the impacts of early disadvantage upon latter educational attainment remains uncertain in both the Educational Effectiveness and Risk and Resilience research traditions. Here, both fields are drawn upon in a prospective longitudinal investigation of 2,664 children between the ages of 6-11 years whose academic skills in English and maths along with self regulation were measured at ages 6, 7, and 11 years( based on the Effective Provision of Pre-school and Primary Education 3-11 project sample). The results of a multilevel Structural Equation model that allowed us to test the hypothesis that attending a more academically effective primary school could lessen the adverse developmental impact of experiencing multiple (early) disadvantages on child outcomes measured over-time at ages 6, 7, and 11 years will be presented. Experiencing a greater number of early disadvantages between birth to age 5 was found to strongly impair self regulation and academic attainment throughout primary school. However, attending a more academically effective primary school for just a single year was found to partially protect reading, maths, and self regulation outcomes at age 6 from the adverse impact of early disadvantage. Further, more academically effective primary schools were also found to offer an additional longer-term form of protection - they significantly lessened the extent to which earlier abilities in reading, writing, and self regulation predicted these same abilities at age 11. Although more academically effective primary schools cannot remove the impacts of disadvantage, the results suggest they can make a significant and positive difference to the longer term academic attainment and self regulation of primary school children who experienced more disadvantages before the start of school and so help to mitigate their negative consequences.

Two- and three-level structural equation models

Dr Lars-Erik Malmberg, Department of Education

13 February 2012 12:15 - 14:00
Seminar Room J

Conveners: Professor Pam Sammons, Dr Lars-Erik Malmberg and Dr James Hall

Interaction, Moderation, and Mediation: Definitions, Discrimination, and (some) means of testing

Dr. James Hall, Department of Education

06 February 2012 12:15 - 14:00
Seminar Room J

Conveners: Professor Pam Sammons, Dr Lars-Erik Malmberg and Dr James Hall

Abstract
In 1986 Baron and Kenny set out to clarify the differences between the terms “Moderation” and “Mediation” as used in the social sciences. Twenty six years later, the seminal paper that this collaboration resulted in (published in the Journal of Personality and Social Psychology) has been around 30,347 times (Google Scholar on 12/01/2012). This is an average of 1,167 each year - the equivalent of more than once every 8 hours for over a quarter of a century. Despite this citation record, the uncertainty surrounding these terms has not gone away. Academics still struggle to define, distinguish and utilise these terms while related undergraduate teaching is still an exception. This presentation sets out simple, clear definitions that distinguish “Interaction”, “Moderation”, and “Mediation” as well as a number of other commonly-used terms. An introduction is given on how to use these concepts in ‘real-life research’ with worked-through examples provided. The presentation slides themselves also serve as a short primer for future reference.

Measurement Error and School Effects

John Fletcher, Department of Education

30 January 2012 12:15 - 14:00
Seminar Room J

Conveners: Professor Pam Sammons, Dr Lars-Erik Malmberg and Dr James Hall

Abstract
Historically, school effectiveness has been estimated using multi-level models, which have not taken account of measurement error. Failure to take account of measurement error leads to biased estimation. In this case study, I investigated whether taking account of measurement error in the English Contextual Value Added model leads to substantively important differences in the overall size of school effects or in the ranking of schools. I took account of measurement error on National Curriculum end stage tests used in this model by comparing pupils' scores on pairs of similar tests. I tested the sensitivity of the model to measurement error on eligibility for free school meals (FSM) by fitting hypothetical error variances and by incorporating a contextual aggregate in the model. I found that taking account of measurement error has little impact on the estimation of school effectiveness in this model. The inclusion of a contextual aggregate for FSM marginally reduced the estimated variation in the contribution of schools.

Schooling, Child Labour and School Quality

Professor James Foreman-Peck, University of Cardiff

23 January 2012 12:15 - 14:00
Seminar Room J

Quant SIG seminar

Professor Herb Marsh, Department of Education

28 November 2011 12:15 - 14:00
Seminar Room J, 28 Norham Gardens

Part of the Michaelmas Term 2011 Quant SIG Programme

Abstract
Forthcoming

League tables and school choice

Dr. Harvey Goldstein

21 November 2011 12:15 - 14:00
Seminar Room J, 28 Norham Gardens

Part of the Michaelmas Term 2011 Quant SIG Programme

Abstract
The use of value added league tables in education is widespread. A major justification is for choosing schools by parents. This presents both technical and conceptual problems. In this paper we use the National pupil database to explore these issues and make recommendations.

Testing non-nested regression models and distinguishing additive from multiplicative statistical interactions

Dr. James Hall, University of Oxford, Department of Education & University of Warwick, Department of Psychology

14 November 2011 12:15 - 14:00
Seminar Room J, 28 Norham Gardens

Co-author: Julia Chernova University of Warwick, Department of Psychology

Part of the Michaelmas Term 2011 Quant SIG Programme

Abstract
The difference-testing of 2+ statistical regression relationships relies upon comparing a “full” statistical model to an otherwise identical but “restricted” or “nested” model in which the 2+ statistical relationships are fixed to equality. The specification of this nested model provides a null-hypothesis model (where the relationships are equal to one another). There are however incidents where difference-testing is required but the specification of a nested model is either difficult or impossible. One such incidence is when alterative measures of the same concept need to have their parallel effects compared as they impact some outcome. Here we consider the less well-known statistical tests that compare non-nested models and consider a ‘real-life’ example in which these have been used: comparing two competing hypotheses for how a set of measures may work-together/interact.

Understanding student aspirations: enhancing the learning environment

Dr. Russell J. Quaglia, President/Founder, Quaglia Institute for Student Aspirations

07 November 2011 12:15 - 14:00
Seminar Room J, 28 Norham Gardens

Part of the Michaelmas Term 2011 Quant SIG Programme

Abstract
This presentation will describe the dynamic nature of student aspirations and advances an integrative and cohesive definition of aspirations as identity-relevant, future-oriented goals which one strives in the present to actualize. Furthermore, we posit a process model whereby three school-related constructs—teacher support, positive peer environment, and school engagement—promote aspirations. Specifically, we believe that teacher support and a positive peer environment predict greater engagement, which in turn promotes higher levels of aspirations. Dr. Quaglia will share data that has been gathered from over 500,000 middle and high school students that depicts their perceptions of their teaching and learning environment. His presentation will have implications regarding how schools are organized, the role of educators, and how we are assessing students. The presentation will also provide an understanding how student aspirations can be used proactively in school reform initiatives designed to increase academic standards and in the overall develop of healthy productive citizens.

Overcoming non-random assignment to treatment conditions in observational studies: an introduction to Propensity Score Analysis

Dr. Carlo Chiorri, Visiting Scholar, Department of Education

31 October 2011 12:15 - 14:00
Seminar Room J, 28 Norham Gardens

Part of the Michaelmas Term 2011 Quant SIG Programme

Abstract
True experimental design (aka randomised control trial, RCT) is regarded as the most accurate form of experimental research, and it is often thought of as the only research method that allows an adequate assessment of causal effects. One of its main features is random assignment to treatment conditions, which ensures that the group attributes for the different treatments will be (roughly) equivalent and therefore any effect observed between treatment groups can be ascribed to the treatment effect and not to pre-existing differences of the individuals in the groups. However, in education and in social sciences, RCTs are not always possible, feasible, or even ethical, and often researchers need to assess treatment effects from observational data, i.e. data collected through observation of systems without any attempt to manipulate independent variables or to randomly assign cases without treatment conditions. In the last three decades social science researchers have developed new, efficient approaches for controlling for covariates and assessing treatment effects from studies based on observational data. Substantial contributions came from statisticians and econometricians, and collectively the new approaches are known as Propensity Score Analysis (PSA). This contribution aims at introducing the basic features of PSA, its potential applications and the R packages that can be used to perform the analysis.

Using the polytomous Rasch model to equate two tests which measure the same proficiency

Professor David Andrich, Chapple Professor, Graduate School of Education, University of Western Australia

24 October 2011 12:15 - 14:00
Seminar Room J, 28 Norham Gardens

Part of the Michaelmas Term 2011 Quant SIG Programme

Abstract
Educational test equating is an integral task of modern psychometrics. It is used in high stakes, university entrance examinations in Australia to equate tests from different discipline areas. This paper describes equating using the polytomous Rasch model in which the person parameters are conditioned out, making the equating independent of the distribution of persons. This contrasts with most other methods of test equating.

Peer Effects in Academic Achievement: An Approach via Network Dynamics

Professor Tom Snijders, Department of Politics and Department of Statistics, University of Oxford

17 October 2011 12:15 - 14:00
Seminar Room J, 28 Norham Gardens

Part of the Michaelmas Term 2011 Quant SIG Programme

Abstract
Studies of peer effects in educational settings confront two main problems. The first is the presence of endogenous sorting which confounds the effects of social influence and social selection on individual attainment. The second is how to account for the local network dependencies through which peer effects influence individual behaviour. These problems are addressed here using a longitudinal data set on academic performance, friendship, and advice seeking relations among students in a full-time graduate academic program. Stochastic actor-based models are specified that permit estimation of the interdependent contribution of social selection and social influence to individual performance. In this data there is evidence of peer effects, i.e., students tend to assimilate the average performance of their friends and of their advisors. At the same time, students attaining similar levels of academic performance are more likely to develop friendship and advice ties. Together, these results imply that processes of social influence and social selection are sub-components of a more general co-evolutionary process linking network structure and individual behaviour. The approach followed here is discussed in the wider frame of issues of causality in the study of peer effects. This presentation is based on joint work with Alessandro Lomi, Vanina Torlò, and Christian Steglich.

School achievement for immigrant and local students: exploring the predictive role of problem solving in Programme for International Student Assessment (PISA) data

Andrew Martin, University of Sydney

13 June 2011 12:15 - 13:15
Seminar Room J

Quant SIG Seminar Series
Convener: Professor Herb Marsh Director, SELF Research Centre

Investigating the Big Fish Little Pond Effect with English Primary School attainment data

Ioulia Televantou: University of Oxford, SELF DPhil Student

06 June 2011 12:15 - 13:15
Seminar Room J

Quant SIG Seminar Series
Convener: Professor Herb Marsh Director, SELF Research Centre

An introduction to a new class of multilevel models – termed hierarchical related regressions (HRR) – for estimating individual-level associations using a combination of aggregate (group level) and individual-level data

Steven Fisher, University of Oxford, Department of Sociology

23 May 2011 12:15 - 13:15
Seminar Room J

Quant SIG Seminar Series
Convener: Professor Herb Marsh Director, SELF Research Centre

The curious case of Big Five questionnaires confirmatory factor analyses: issues and solutions

Carlo Chiorri, University of Genova, Department of Anthropological Sciences, Psychology Unit

16 May 2011 12:15 - 13:15
Seminar Room J

Quant SIG Seminar Series
Convener: Professor Herb Marsh Director, SELF Research Centre

The lack of teacher gender effect on boy and girl students: primary and secondary effects on entering university

Martin Neugebauer, University of Mannheim

09 May 2011 12:15 - 13:15
Seminar Room J

Quant SIG Seminar Series
Convener: Professor Herb Marsh Director, SELF Research Centre

Quant SIG meeting

07 March 2011 12:15 -
Seminar Room J, 28 Norham Gardens

Convener: Herb Marsh

Quant SIG meeting

28 February 2011 12:15 -
Seminar Room J, 28 Norham Gardens

Convener: Herb Marsh

Quant SIG meeting

John Fletcher, Department of Education, University of Oxford

21 February 2011 12:15 -
Seminar Room J, 28 Norham Gardens

Convener: Herb Marsh

Bullies and victims: measurement, psychosocial determinants, intervention, and the role of self-concept (Public Seminar)

Professor Herb Marsh, Department of Education

14 February 2011 17:00 - 18:30
Seminar Room A

Convener: Kathy Sylva

Abstract
Bullying is the systematic prolonged abuse by other generally more powerful groups or individuals. It incorporates a wide range of behaviours, such as, name-calling, physical violence, exclusion, and verbal and physical intimidation. Both engaging in and being the target of bullying are significant risk factors that threaten long term psychological and personal development.

The present investigation had three overaching aims: First, it sought to create sound psychometric multifactorial instruments for secondary students to reliably measure the nature and frequency of bullying incidents, bystander roles, school climate, and related outcomes. Secondly, this study aimed to explicate the psychosocial determinants of bullying, including the role of self-concept, to identify characteristics which differentiate bullies and their targets, and explicating the role of other motivators in the incidence/maintenance of bullying and being bullied. Finally it aimed to comprehensively evaluate the effectiveness of a new teacher centred fully standardised and manualised whole school anti-bullying intervention, the Beyond Bullying Secondary Schools Program, on reducing the frequency of bullying incidents and enhacing school climate among other factors.

A total of 3522 secondary school students (1510 males & 2012 females) in grades 7 to 11 and 256 teachers across 6 secondary non-government schools in NSW, Australia participated in this study. Confirmatory factor analysis, reliability analysis and structural equation models of factorial invariance were used to assess the hypothesised factor structure and psychometric properties of the instruments used in this study. Structural equation models were used to analyse both cross-sectional, mean differences and longitudinal causal models. to determine the causal nature of bullying on such factors as depression and low self-concepts. The psychometric evaluation of the measures in this study are the among the most robust to be conducted in any bullying research study to date.

Multilevel modelling was used to evaluate the effectiveness of the Beyond Bullying Program on desired outcomes such as the reduction in overall bullying rates. This study further supports the need for whole school interventions but extends the findings of previous research by showing that standardised, manualised interventions across schools which concentrate not only on whole school activities but also on specifically training teachers with skills to manage bullying incidents they withness can greatly enhance their efficacy in reducing bullying in secondary schools.

Primary school students' self-regulation during ill- and well-structured tasks: A dual growth model approach

Dr Lars Malmberg & Jason Smart, Department of Education, University of Oxford

14 February 2011 12:15 -
Seminar Room H

Convener: Herb Marsh

The route out of the routine: mobility and the changing structure of occupations

Dr. Craig Holmes, Department of Education, University of Oxford

07 February 2011 12:15 -
Seminar Room J, 28 Norham Gardens

Convener: Herb Marsh

Propensity score measurement

Dr. Benjamin Nagengast, Department of Education, University of Oxford

31 January 2011 12:15 -
Seminar Room J, 28 Norham Gardens

Convener: Herb Marsh

The influence of child, family, home factors and pre-school education on the identification of special educational needs at age 10.

Professor Pam Sammons, Department of Education, University of Oxford

24 January 2011 12:15 -
Seminar Room J, 28 Norham Gardens

Convener: Herb Marsh

The early identification of young children’s special educational needs (SEN), as well as the development of specific strategies to support those children identified with special needs, are increasingly recognised as crucial to facilitating good adjustment to school and to ensuring that such children are helped to reach their full potential in education. Using a large national sample of young children in England whose developmental progress was followed up from pre-school, this study investigates which child, family, home and pre-school factors can be viewed as risk or protective factors for later SEN-status at age 10. The experience of high-quality pre-school education is shown to reduce the likelihood of a child being identified as experiencing SEN in the long run. Teachers’ assessments of SEN are found to be strongly related to children’s reading and mathematics attainment, but other factors also predict SEN, including a child’s age within a year group.

(Yvonne Anders, Pam Sammons, Brenda Taggart, Kathy Sylva, Edward Melhuish and Iram Siraj-Blatchford (2010) - The influence of child, family, home factors and pre-school education on the identification of special educational needs at age 10. British Educational Research Journal 2010, 1–21)

Studying the Phantom Effect Using PIPS data

Speaker: Ioulia Televantou Department of Education, University of Oxford

29 November 2010 12:15 -
Seminar Room J



SEM & Latent Growth Modelling across different measures: one instance of defensible validity

22 November 2010 12:15 -
Seminar Room J

Speaker: James Hall Department of Education

Within-teacher stability and variability: An unexplored dimension of teachers’ self-efficacy development

15 November 2010 12:15 -
Seminar Room J

Speaker: Lars-Erik Malmberg Department of Education

Teachers who think they have what it takes to support students’ learning have a sense of competence, confidence, agency, or efficacy. In 1984, Gibson and Dembo set the tone for a quarter of a century of research on teacher self-efficacy by summarizing advancements in the field. They highlighted seven areas which needed further investigation. In brief, these were adding further theoretical aspects; validating measures; teasing apart teacher characteristics from organisational and situational aspects of teaching; and inspecting the associations between self-efficacy and teaching style, classroom process variables, learning outcomes, and classroom management. Twenty-four years later Dellinger, Bobbett, Olivier, and Ellett (2008) stated that advancements had been made to address several of the areas pointed out by Gibson and Dembo (1984), but reiterated the need for further research.

The first aim of my presentation is to summarize recent advancements in the field. The second aim is to focus on research in areas that appear under-investigated to date, namely (a) the distinction between person- and situation-specific characteristics of teachers’ beliefs, perceptions and behaviours, and (b) the distinction between longer-term teacher development and shorter-term stability and variability in situation-specific (intrapersonal) beliefs, perceptions and behaviour.

In order to distinguish between interpersonal (differences between persons) and intrapersonal (differences within persons) aspects of teachers’ beliefs, perceptions and behaviours I elaborate on the heuristic framework for conceptual analysis of agents (teachers), means (domains of teacher competence), and goals (student learning or performance) by Ellen Skinner (1996). Within this framework, I review findings from studies of longer-term changes in teachers’ beliefs, perceptions and observed behaviour during teacher education and during the transition to working life (Malmberg, 2008; Malmberg & Hagger, 2009a; Malmberg et al., 2009). I contrast these findings with results from shorter-term studies of teachers’ intrapersonal beliefs (e.g., the Teaching Every Lesson study; Malmberg & Hagger, 2009b): competence enactment and competence evaluation, perceptions of student engagement, and affect.

Teachers varied in their competence evaluation and perceptions of student disengagement from one lesson to the next, and from one classroom to another, suggesting (in line with Raudenbush, Rowan, & Cheong, 1992) that student group characteristics indeed account for part of the variance in teachers’ competence beliefs. Teacher perceptions of student disengagement from one lesson to the other, also related to negative affect and interpersonal characteristics such as exhaustion.

The talk concludes with an outline for further research on teachers’ interpersonal and intrapersonal beliefs, perceptions and behaviour, and particularly on how such studies could be designed. Moreover, I outline how modern technology can facilitate new methods of data collection, how cutting edge statistical techniques can be used for testing of quality of measurement and modelling of substantive results, and how advances in our knowledge concerning teacher development can provide insights into the challenges teachers are face in the 21st Century.

Social comparison with friends versus non-friends

08 November 2010 12:15 -
Seminar Room J

Speaker: Miranda Lubbers Department of Social and Cultural Anthropology, University of Barcelona

There is an interesting analogy between two areas of research that address the influence of specific classmates on a pupil´s performance, namely social comparison research and peer socialisation research. In this presentation, I will focus on the interface between these two areas. Specifically, I will discuss to what extent pupils in secondary education compare their grades with those of their friends, whether several aspects of social comparison differ as a function of the type of relation between a pupil and his or her target, and whether individual characteristics affect the choice of friends versus non-friends as comparison targets. Participants were 9612 pupils in the first grade of secondary education in the Netherlands (equivalent to Year 8 in England). Results indicated, among others, that the vast majority of pupils compared their grades with those of friends, that the tendency to compare with friends varied with gender and personality, and that the choice of friends was associated with several differences in the further process of social comparison. Despite these differences, which suggested that friends often serve as routine standards, consequences of comparison for subsequent performance were about the same for both types of relations. The results further suggested that social comparison theory provides an explanation for previously found effects of friends’ influence on school performance. The presentation will conclude with a discussion of future research.

The stability of school performance indicators: comparing cross-sectional and longitudinal models

01 November 2010 12:15 -
Seminar Room J

Speaker: Xavier Dumay Interdisciplinary Research Group in Socialisation, Education and Training, Université Catholique de Louvain, Belgium

In the literature on school effectiveness, little work has concentrated on the stability of school performance over time. The goal of this paper is to examine the role played by the approach used in estimating school performance on this parameter. Using a multi-cohort longitudinal database (five cohorts, from 1999 to 2004), four models of school performance estimation are compared. The first one is a null model, with school performance defined as the average performance of students at the end of grade 5. In the second model of estimation, the average performance of students at the end of grade 5 is controlled for their individual performance at the end of grade 1 in a multilevel model. In the third and the fourth model, multilevel growth models (MGM) are used. In model 3, only the (linear) growth parameter is defined as random, with school time residuals being considered as school performance indicators. In model 4, intercept and slope are considered as random. The results show that 1) the average stability over time is significantly influenced by the level of schools’ scores adjustment (unadjusted vs. adjusted) and the model type (cross-sectional vs. growth), and tends to be affected by their interaction, and that 2) the reliability of schools’ performance and their dependence on the schools’ social composition score is less when scores are adjusted, and in particular for growth scores.

A multifoci person-centered perspective on workplace affective commitment: a Latent Profile/Factor Mixture Analysis

25 October 2010 12:15 -
Seminar Room J

Speaker: Alexandre Morin University of Sherbrooke, Canada

The present study aims to explore the usefulness of a person-centered perspective to the study of workplace affective commitment (WAC). Five distinct profiles of employees were hypothesized based on their levels of WAC directed toward seven foci (organization, workgroup, supervisor, customers, job, work and career). This study applied latent profile analyses and factor mixture analyses to a sample of 404 Canadian workers. The construct validity of the extracted latent profiles was verified by their associations with multiple predictors (gender, age, tenure, social relationships at work, workplace satisfaction and organizational justice perceptions) and outcomes (in-role performance, organizational citizenship behaviors and intent to quit). The analyses confirmed that a model with five latent profiles adequately represented the data: (a) highly committed toward all foci; (b) weakly committed toward all foci; (c) committed to their supervisor and moderately committed to the other foci; (d) committed to their career and moderately uncommitted to the other foci; (e) committed mostly to their proximal work environment. These latent profiles present theoretically coherent patterns of associations with the predictors and outcomes, which suggests their adequate construct validity.


Quant Sig Meeting

18 October 2010 12:15 -
Seminar Room J

Title and speaker to be confirmed

Switching on and switching off in mathematics: a longitudinal and multilevel study of engagement amongst middle school students

11 October 2010 12:15 -
Seminar Room J

Speaker: Andrew Martin Faculty of Education and Social Work, University of Sydney

Authors: Andrew J. Martin, Janette Bobis, Judy Anderson, Jennifer Way, and Rosemary Vellar University of Sydney.