Research, Data & Advocacy Lab

students learn best when they are exploring ideas they genuinely care about — and when they are given the tools to express those ideas with confidence

  • Intellectual confidence

  • Motivation driven by curiosity

  • Ownership over their learning

  • Comfort expressing complex ideas

Fostering self efficacy in students, this program is designed to help students develop:

This lab designed to give students the time and structure to deeply explore a research topic of their choice. With expert guidance, students learn how to analyze complex ideas, build evidence, and communicate their thinking with clarity and confidence. By the end of the lab, students produce a polished research paper or participate in a culminating experience that demonstrates their advocacy and communication skills.

A full program length is usually 12 weeks, with weekly meetings for courses, check-ins, and discussion. Students have the freedom to continue after program completion if they are looking for a more extensive result such as participating in publication journals or competitions. This program is suitable for a wide range of students from late middle school, high school, and undergraduates.

Check out some of our student work

  • Paper by Zhihao (CS major)

    Synopsis:
    Peer review is of great importance as it is the cornerstone of academic integrity. It guarantees that published research satisfies certain standards regarding methodology, ethics, and reputation, while also addressing concerns about potential biases. Moreover, fairness in peer review, as shown in studies like \citet{zhang-etal-2023-causal-matching}, is crucial. This involves an analysis of how elements such as author identity and institutional reputation can affect review results, thus sparking more interest in algorithmic fairness in peer review systems.And the growing volume of academic paper submissions has burdened the peer review process, leading to the creation of tools like ReviewRobot \citet{wang-etal-2020-reviewrobot} that use NLP and machine learning to enhance review efficiency. Furthermore, research by \citet{gao-etal-2019-rebuttal} on the ACL - 2018 review process reveals that the rebuttal stage can greatly influence review outcomes, pointing to potential areas for process improvement. Finally, this research employs linear regression, logistic regression, and sentiment analysis to comprehensively understand the various factors influencing review outcomes in the peer review process, thereby contributing to a more effective system. Each of these points provides a foundation for discussing the research's motivations, background, and methods, helping to establish the study's objectives and relevance.

  • Paper by Aley (Psychology major)

    Synopsis:
    Dreams are a bridge between conscious and unconscious experiences, which impact emotional regulation and memory consolidation. The current study investigates if daily life experiences—specifically, movie stimuli—affect dream content and emotionality. By leveraging natural language processing (NLP) and deep learning, we contrast semantic similarity between participants' dream reports and the movies they watched before bedtime. The Dream Emotion Assessment Dataset (DEAD), which contains dream reports, EEG activity, and emotional ratings, provides a multi-modal framework for the study of such associations. The results indicate that, while some elements of movies appear in dreams, no strong correlation between movie-evoked emotions and content of dreams exists. Complexity of dreams, measured in sentence and word length, fails to strongly predict emotional intensity.These results challenge traditional understanding of dream continuity and highlight the nonlinearity of unconscious processing. Future research can advance computational models, investigate a range of real-world stimuli, and look into how dreams assimilate external experience. Results gained can enhance psychological therapy, regulation of emotions, and cognitive neuroscience.

    • Keywords: Dreams，Emotion Regulation in Dreams，EEG-Based Dream Analysis，Natural Language Processing (NLP) in Dream Analysis

  • Paper by Priscilla (Economics major)

    Synopsis:
    Nowadays, online platforms  such as Yelp, Zomato, and Safegraph generate massive datasets that provides new opportunities to study the restaurant industry. These datasets not only capture how users rate restaurants, but also reflect how policies like minimum wage changes affect business operations. In addition, consumer preferences play a key role in shaping dining decisions, including distance and convenience. The advanced model applied to these datasets can help reduce food waste and demand prediction, while external shocks like COVID-19 have reshaped consumer behavior. The purpose of this paper is to apply data-driven methods to lead people to better understand restaurant rating, consumer demand, and the decision-making that improve industry outcomes.

  • Paper by Highschooler

    Abstract:
    https://github.com/jasminerliu/SHTEM-educational-AI/blob/ecb1474dd23a21af5a0b227e15a791a3d587d319/SHTEM_2022.pdf

  • Paper by Abby (Gender Studies Major)

    Synopsis:
    Gender stereotypes have always been an issue in achieving gender equality in society. It can occur in social norms, political policies, or even psychological aspects. From a psychological perspective, one of the stereotypes that often comes to people’s minds is the assumption that “women are more emotional, and men are more rational.” Females are often being assumed to display more emotions and feelings. On the other hand, society expected males to be able to conceal their feelings in order to maintain the image of masculinity. Due to that, certain emotions are more likely to have an association with genders. For instance, emotions like happiness, sadness, and guilt are often socially constructed as "feminine"; emotions like aggression, seriousness, and protectiveness are often connected with men’s emotional expression. This paper will look into whether certain emotions are more likely to be associated with one gender compared to the other in the Emotion-Gendered Stereotypes dataset. In the process of developing my research topic and questions, I applied several methods, including Exploratory Data Analysis to identify patterns, Linear Regression and Logistic Regression to test relationships between variables, and K-Means Clustering to explore patterns in the dataset and to guide the formulation.