Quantile regression is a powerful tool for detecting exposure-outcome associations given covariates across different parts of the outcome's distribution, but has two major limitations when the aim is to infer the effect of an exposure. Firstly, the exposure coefficient estimator may not converge to a meaningful quantity when the model is misspecified, and secondly, variable selection methods may induce bias and excess uncertainty, rendering inferences biased and overly optimistic. In this paper, we address these issues via partially linear quantile regression models which parametrize the conditional association of interest, but do not restrict the association with other covariates in the model. We propose consistent estimators for the unknown model parameter by mapping it onto a nonparametric main effect estimand that captures the (conditional) association of interest even when the quantile model is misspecified. This estimand is estimated using the efficient influence function under the nonparametric model, allowing for the incorporation of data-adaptive procedures such as variable selection and machine learning. Our approach provides a flexible and reliable method for detecting associations that is robust to model misspecification and excess uncertainty induced by variable selection methods. The proposal is illustrated using simulation studies and data on annual health care costs associated with excess body weight.
Large Language Models (LLM) have experienced strong development in recent years, with varied applications. This paper uses LLMs to develop a post-hoc process that provides more elaborated explanations of the results of food recommendation systems. By combining LLM with a hybrid extraction of key variables using SHAP, we obtain dynamic, convincing and more comprehensive explanations to lay user, compared to those in the literature. This approach enhances user trust and transparency by making complex recommendation outcomes easier to understand for a lay user.
Research summary We assess how personality alignment in investor-founder dyads is associated with the likelihood of follow-on funding, a vital outcome for early-stage ventures. Using machine learning to infer the Big Five personality traits from Twitter data for 9497 business angel-founder dyads, we find that similarity in conscientiousness and agreeableness is associated with a higher likelihood of follow-on funding, while similarity in neuroticism is associated with a lower likelihood. We attribute these patterns to the trait-specific benefits of supplementary (conscientiousness and agreeableness) and complementary (neuroticism) fit. Robustness checks and additional analyses support and nuance the conclusion that personality fit matters for venture outcomes, highlighting the strategic role of personality fit in the investor-founder relationship.Managerial summary We show that personality similarity between business angels and founders is associated with whether a venture secures follow-on funding. Similarity in conscientiousness (being organized and reliable) and agreeableness (being cooperative and trusting) is linked to a higher likelihood of securing follow-on funding, while similarity in neuroticism (emotional instability) is linked to a lower likelihood. We interpret these patterns as collaboration dynamics: similarity can help through supplementary fit (e.g., shared work style and cooperation), but differences can help through complementary fit (e.g., one partner's emotional stability offsets the other's emotional instability). Practically, founders and business angels should develop self-awareness and consider personality fit alongside other characteristics when forming partnerships. Policymakers and incubators can support better matches and strengthen collaboration by promoting awareness of interpersonal dynamics.
The international shift towards inclusive education has increased the enrolment of autistic students in mainstream schools. While mainstream education provides autistic students with valuable opportunities for their (social) development, research consistently shows that these environments struggle to fully include autistic students and address their needs. Taking self-determination theory as a guiding framework for assessing students' basic psychological needs and how these are met by teachers, this study used a joint display analysis to combine different streams of data from secondary school autistic students (N = 13; 6 Dutch, 7 Mexican): (1) video observations of classroom interactions analysed using a coding scheme based on self-determination theory, (2) questionnaires about their perspectives on their lessons and interactions with teachers and (3) students' perspectives on these observations obtained through video-stimulated recall interviews. Using joint display analyses, we compared the data from these three sources at both the individual and group level to identify patterns of convergence (agreement), complementarity (expansion) and divergence (contradictions). In doing so, we present a fine-grained, multifaceted picture of the needs of autistic students in mainstream secondary schools and how these are met, incorporating the students' own interpretations into the analysis.
A strong link exists between spatial and numerical abilities, but establishing causal relations remains challenging due to limited longitudinal research and mixed findings from training studies. There is also little consensus on which spatial subdomains are most predictive at different developmental stages, as few studies have directly compared multiple spatial abilities or distinguished between numerical outcomes while controlling for prior skills. To address this gap, the present longitudinal study followed 148 children to examine how three spatial subdomains-intrinsic-static, intrinsic-dynamic, and extrinsic-static abilities-assessed in kindergarten, predict addition and subtraction performance in first grade. Analyses controlled for sociodemographic factors, phonological awareness, and prior numerical knowledge. Hierarchical multiple regression revealed that spatial abilities, along with backward counting, significantly predicted first-grade arithmetic skills. Specifically, intrinsic-dynamic ability was associated with addition, whereas extrinsic-static ability predicted subtraction. Nonsymbolic and symbolic magnitude comparisons, phonological awareness, and sociodemographic variables were not significant predictors when spatial abilities were included. These findings underscore the general importance of spatial skills in early arithmetic development and highlight their role as scaffolds for newly acquired skills. Furthermore, they demonstrate qualitative differences in how specific spatial subdomains support addition versus subtraction, providing actionable insights for designing targeted kindergarten interventions aimed at enhancing foundational mathematical abilities.
