Applied Research

I am a cognitive scientist dedicated to answering questions about human behavior with methodological and analytical rigor, promoting an inclusive community that grows potential for all learners, and learning new skills and frameworks.

I value open science in my behavioral research design, analysis, and online programming and post my experimental and analytic code (Python, R, Matlab,  JASP, JavaScript, Qualtrics) to my Github or OSF (see example repository for in press JEP:HPP paper and example repository for in press Psych Science paper).

I value good science communication. As the instructor of record for Introduction to Cognitive Psychology, I created a Github page for the class and received near perfect course evaluations. I mentored researchers through running participants in crowdsourced experiments and coding in JavaScript. I led and earned a Professional Development Grant from the Graduate School for my Department to create more inclusive programming. I also supervised and edited summaries of Duke research (e.g., 1, 2) and blogged about trans* inclusive pedagogy and the value of writing groups.

For some select examples of applied projects, see below.

 

dibs-germinator-tw-v2
Read the official announcement of the Germinator grants.

Do individual differences in attentional states relate to individual differences in mental health? 

JavaScript, HTML/CSS, Qualtrics, R, Confirmatory Factor Analysis, Computational Modeling, Experiment Design, Big Data, Usability Testing, Regression

Problem: Studies on patients with psychiatric diagnoses typically involve small numbers of patients and use behavioral surveys to assess cognitive deficits that don’t relate to the constructs as precisely as cognitive psychologists would hope. Meanwhile, attention researchers typically ignore clinical research when thinking about how to advance basic science theories. Creating an interdisciplinary bridge between the fields, I successfully earned a grant to run two large (>2k participants) online crowdsourced experiments, with participants performing attention experiments and self-reporting psychiatric symptoms.

Process: After earning the grant, I wrote a draft of a preregistration proposal, stating hypotheses, proposed experimental design, analyses, etc. and received feedback from my collaborators. Since then, I have been piloting the best approach to collect the data. For example, I have run side projects to ensure the attention experiments and self-report psychiatric questionnaires are intuitive for our participants (e.g., instructions make sense, code works well) and replicate the results we are expecting. I have consulted with my collaborator on possible reinforcement learning models that we will use when analyzing the data.

Outcome: In progress. My research adviser incorporated a form of this proposal in his R01 grant, “Neural mechanisms of cognitive meta-flexibility,” which was funded. I anticipate at least two research publications from this work: one focused on relating individual differences in attentional states (via reinforcement learning / computational modeling) to mental health (via confirmatory factor analysis) and one focused on providing supporting evidence for the role of learning in attention (via confirmatory factor analysis). This project will likely be a chapter of my dissertation.

 

artshow
Art Exhibit on this research in the Perkins Library at Duke

What are the constellation of learning beliefs that students hold that impact academic success, motivation, and well-being?

Qualtrics, Survey Design, Big Data, Log Analysis, Web Application Development, Prototyping, Team Management, A/B Testing

Problem: Whether people believe intelligence cannot be developed (fixed mindset) or can be developed through effort and experience (growth mindset) has profound effects on academic success. However, most mindset research has focused exclusively on predicting academic grades to the detriment of understanding student well-being, grade satisfaction, and other naturalistic behaviors in the classroom. Moreover, intelligence mindset is currently measured with a 4-item scale that is quite explicit in what is being tested. As intelligence mindset as a concept has become more popular (e.g., >2 million copies of a popular science book sold), this measurement validity issue has likely posed problems (e.g., socially desirable answers, etc.).

Process: My collaborator and I obtained funding through the Charles LaFitte Foundation. We recruited instructors from math, stats, computer science, psychology, chemistry, and physics classrooms for our research and sent out Qualtrics surveys assessing student beliefs. To measure mindset, we used both the scale and open-ended prompts (e.g., What caused you to have your current level of intelligence?). We trained research assistants to code free-text responses to these prompts as reflective of the intelligence mindsets. We also have extracted event logs from learning management sites (Sakai, Piazza) to formalize help-seeking and procrastination-related behaviors.

Outcome: We have a paper currently under review at Teaching of Psychology on a data-driven approach to teaching intelligence mindsets. We created a web-based algorithm categorizing the language students use as indicative of intelligence mindsets. We curated an art exhibit of this research to challenge the way we think about intelligence. We presented the preliminary results of our project at the BRITE Ideas talk series in 2018 and 2019. Further outcomes are anticipated: we have over 2,500 responses to analyze and data from over thirteen STEM classrooms. We also have collaborators implementing these surveys outside of Duke.

 

litlushlogo

Could we create a modern, user-centered social media site for book lovers?

Prototyping, Business Management, Web Development, Recruitment, Social Media Start-ups

Problem: In 2013, my business partner and I thought that Goodreads, the predominant social media site for readers, had an outdated design and was missing key demographics of the book community. For example, Goodreads does not allow users to display additional videos or images, forcing users to link off-site and missing many different book communities – booktubers, i.e., Youtubers who make videos about books; bookstagrammers, i.e., Instagrammers who photograph books; etc. LitLush, pitched as social media for book lovers where Youtube meets Goodreads, was born.

Process: My business partner/the CEO recruited a manager who subsequently reached out to web developers. I recruited beta users through my connections in the book community; we specifically aimed for users whose reach and follower counts were high and who constituted our target audience. The CEO and I strategized together about the logo – the use of gender-neutral colors, modern fonts, and the symbol. We discussed branding: Long Live Lit became our slogan. We provided feedback to the developers on the prototype pages provided to us: member profile page for users, book profile pages, and the video portion of the site (not final).

Outcome: The site got to a testing stage before funding ran out, in part because the developers underbid the estimated amount of work involved. However, all was not lost, because the CEO and I made valuable connections with our beta testers. These connections were translated into the young adult villain-themed booktuber/author anthology, Because You Love to Hate Me, which we both contributed to and was a New York Times bestseller.