Lucas Reymond

As a statistical consultant supporting a research team in language science, I contributed to the analysis of longitudinal data from a study investigating how children acquire vocabulary based on their exposure to different types of speech. The central question was whether adult-directed speech (ADS) has a measurable impact on vocabulary growth.

The dataset included cumulative counts of words used by a child and corresponding daily exposure to child-directed speech (CDS), adult-directed speech (ADS), and TV/electronic noise.

Consulting Responsibilities and Methodological Approach

My role as a consultant involved:

• Clarifying the scientific question: I worked with the researchers to determine whether the focus should be on cumulative vocabulary growth or the daily acquisition of new words, helping them better align their question with an appropriate statistical model.
• Diagnosing data issues: I identified inconsistencies such as unusually high vocabulary gains on days with minimal exposure and proposed preprocessing steps to correct or flag these entries.
• Transforming the outcome: To ensure statistical independence, I reconstructed the response variable by computing the daily number of new words, rather than using the original cumulative counts.
• Selecting the right model: I explained the limitations of Poisson models for this overdispersed count data and justified the use of a negative binomial regression, incorporating ADS, CDS, TV, and day number as predictors.
• Interpreting results for non-specialists: I translated statistical outputs into actionable insights, highlighting that ADS had a significant positive effect on vocabulary acquisition, while CDS had a minor negative effect and TV exposure showed a surprisingly positive association.
• Discussing limitations and future directions: I advised on the challenges of working with data from only one child, the absence of lagged variables, the possibility to use richer contextual information (e.g., word types, interaction quality), and the need to better define the nature of “vocabulary acquisition” in future studies.

This project exemplified the role of statistical consulting as not merely technical support, but as a collaborative process of co-defining research questions, refining methodology, and enabling transparent, reproducible, and insightful analysis. My contribution helped the team derive interpretable results while also improving their experimental design perspective for future work.