Need for Cognition and Academic Achievement: The Importance of Learning Environments

Scientific Conference (2020, March)

Contrasting classical and machine learning approaches in the estimation of value-added scores in large-scale educational data.

Scientific Conference (2020, March)

The Identification of Math and Reading Difficulties in a Multilingual Educational Setting

Scientific Conference (2020, January 24)

Contrasting Classical and Machine Learning Approaches in the Estimation of Value-Added Scores in Large-Scale Educational Data

in Frontiers in Psychology (2020), 11

There is no consensus on which statistical model estimates school value-added (VA) most accurately. To date, the two most common statistical models used for the calculation of VA scores are two classical ... [more ▼]

There is no consensus on which statistical model estimates school value-added (VA) most accurately. To date, the two most common statistical models used for the calculation of VA scores are two classical methods: linear regression and multilevel models. These models have the advantage of being relatively transparent and thus understandable for most researchers and practitioners. However, these statistical models are bound to certain assumptions (e.g., linearity) that might limit their prediction accuracy. Machine learning methods, which have yielded spectacular results in numerous fields, may be a valuable alternative to these classical models. Although big data is not new in general, it is relatively new in the realm of social sciences and education. New types of data require new data analytical approaches. Such techniques have already evolved in fields with a long tradition in crunching big data (e.g., gene technology). The objective of the present paper is to competently apply these “imported” techniques to education data, more precisely VA scores, and assess when and how they can extend or replace the classical psychometrics toolbox. The different models include linear and non-linear methods and extend classical models with the most commonly used machine learning methods (i.e., random forest, neural networks, support vector machines, and boosting). We used representative data of 3,026 students in 153 schools who took part in the standardized achievement tests of the Luxembourg School Monitoring Program in grades 1 and 3. Multilevel models outperformed classical linear and polynomial regressions, as well as different machine learning models. However, it could be observed that across all schools, school VA scores from different model types correlated highly. Yet, the percentage of disagreements as compared to multilevel models was not trivial and real-life implications for individual schools may still be dramatic depending on the model type used. Implications of these results and possible ethical concerns regarding the use of machine learning methods for decision-making in education are discussed. [less ▲]

Dimensional and Social Comparison Effects on Domain-Specific Academic Self-Concepts and Interests with First- and Third-Grade Students

Scientific Conference (2019, November 06)

Academic self-concepts (ASCs) are self-perceptions of one’s own academic abilities. The internal/external frame of reference (I/E) model (Marsh, 1986) explains the formation of domain-specific ASCs ... [more ▼]

Academic self-concepts (ASCs) are self-perceptions of one’s own academic abilities. The internal/external frame of reference (I/E) model (Marsh, 1986) explains the formation of domain-specific ASCs through a combination of social (i.e. comparing one’s achievement in one domain with the achievement of others in the same domain) and dimensional (i.e. comparing one’s achievement in one domain with one’s achievement in another domain) comparisons. This results into positive achievement-self-concept relations within the math and verbal domains, but into negative achievement-self-concept relations across these domains. The generalized internal/external frame of reference (GI/E) model (Möller, Müller-Kalthoff, Helm, Nagy, & Marsh, 2015) extends the I/E model to the formation of other domain-specific academic self-beliefs such as interest. Research on the validity of the (G)I/E model for elementary school children is limited, especially for first-graders. This study examined the associations between verbal and math achievement and corresponding domain-specific self-concepts and interests for first-graders and third-graders. Two fully representative Luxembourgish first-grader cohorts and two fully representative third-graders cohorts (N=21,192) were used. The analyses were based on structural equation modeling. The findings fully supported the (G)I/E model for third-graders: Achievement was positively related to self-concept and interest within matching domains. Negative relations were found between achievement and self-concept and between achievement and interest across domains. For first-graders, achievement was positively related to self-concept and interest within matching domains. However, the majority of cross-domain relations were non-significant, except for the negative path between math achievement and verbal interest. Hence, while the formation of domain-specific ASCs and interests seem to rely on social and dimensional comparisons for third-graders, only social comparisons seem to be in operation for first-graders. Gender and cohort invariance was established for both grade levels. The findings are discussed within the framework of ASC differentiation and dimensional comparison theory applied to elementary school students. [less ▲]

Predicting Academic Success in Early Primary School: A Propensity Score Matching Approach

Scientific Conference (2019, November 06)

Monitoring du système scolaire – Le modèle luxembourgeois (invited talk)

Scientific Conference (2019, October 17)

Value-added models: To what extent do estimates of school effectiveness depend on the selection of covariates?

Scientific Conference (2019, September)

Vers une éducation équitable et de qualité pour les élèves luxembourgeois d’aujourd’hui et de demain (invited talk)

Scientific Conference (2019, September)

Assimilation and Contrast Effects of Dimensional Comparisons in Self-Concepts, Interests & Anxieties

Scientific Conference (2019, August 16)

Research on the internal/external frame of reference (I/E) model has frequently found contrast effects of dimensional comparisons (i.e. a negative relationship between achievement and self-concept across ... [more ▼]

Research on the internal/external frame of reference (I/E) model has frequently found contrast effects of dimensional comparisons (i.e. a negative relationship between achievement and self-concept across domains) between math and verbal domains. The generalised internal/external frame of reference (GI/E) model extends the I/E model to multiple domains including multiple languages and to other academic self-beliefs and attitudes. When considering multiple languages, achievement-self-concept relations across languages have been found to be either negative (i.e. contrast effect), positive (i.e. assimilation effect), or non-significant. The present study contributes to the ongoing debate concerning the effect of dimensional comparisons among languages by (1) examining dimensional comparisons across two languages and (2) extending the examination to interest and anxiety as outcome variables beyond self-concept. We analysed domain-specific self-concepts, interest, anxieties, and achievement regarding French, German and math in a representative sample (N=5,789) of Luxembourgish ninth-graders. Findings indicated (1) clear contrast effects in the formation of self-concept and interest in German, French and math, and (2) a combination of contrast, assimilation and/or no effects in the formation of anxiety in math, German, and French. With regard to the latter, contrast effects were found for achievement-anxiety paths from German to French, French to German, and French to math. Achievement-anxiety paths from math to French and German to math were non-significant, while the path from math achievement to German anxiety showed a small, yet significant assimilation effect. Results are contextualised within the multilingual Luxembourgish educational system and implications for research on dimensional comparisons are discussed. [less ▲]