Richard Gonzalez
Center Director, Research Center for Group Dynamics, Institute for Social Research
Co-Director, BioSocial Methods Collaborative
Amos N Tversky Collegiate Professor, Psychology and Statistics, LSA
Professor of Integrative Systems and Design, College of Engineering
| E-mail: | Email Richard Gonzalez |
| Address: | Research Center for Group Dynamics Institute for Social Research University of Michigan 426 Thompson Street Ann Arbor, Michigan 48106 |
| Phone: | 734-647-6785 |
Articles in Statistics/Methods
The runner’s bounce and their performance
Extended our model of capturing heterogeneity in ground reaction forces while running to identify bouncing behavior in elite runners.
Developing a new approach to modeling the ground reaction forces in elite human runners
Runners are commonly modeled as spring–mass systems, but the traditional calculations of these models rely on discrete observations during the gait cycle (e.g. maximal vertical force) and simplifying assumptions (e.g. leg length), challenging the predicative capacity and generalizability of observations. We present a method to model runners as spring–mass systems using nonlinear regression (NLR) and the full vertical ground reaction force (vGRF) time series without additional inputs and fewer traditional parameter assumptions.
Estimating a typical path from GPS data
Finding insights from sensor data such as GPS can be tricky. Commuting between home and work may not always follow the same path as some days there are additional stops for errands or alternate routes taken. We propose an algorithm for extracting a typical path from a collection of trips coded by GPS coordinates.
Predicting satisfaction in romantic relationships
We used machine learning to understand which constructs have greater predictive importance for perceived changes in satisfaction since the pandemic began and satisfaction over the prior week.
Comparison of common amplitude metrics in event-related potential analysis
We applied statistical theory to compare several common amplitude metrics for event-related potential analysis of EEG data
Machine learning and the selection of statistical interactions
Machine learning does a great job of selecting variables to include in a predictive model. But it will not always obey some desired properties that we implement in most analytic strategies, such as including main effect versions of a predictor if that predictor is included in the model in the form of an interaction with other variables. We compare several existing algorithms and propose a new one to address this issue.
Testing construct validation
We propose a method to test the structure of a covariance matrix for fit against a hypothesized structure using a permutation approach.
Describing hospital stay trajectories the year prior to sepsis
Used a latent profile analysis to examine clusters of individual patient trajectories of hospitalizations one year prior to sepsis. We used two testing cohorts and validated that these trajectory classes predict mortality 90 days after sepsis.
When emotions are both positive and negative
We propose a new measure for assessing mixed emotions over daily activities in older adults. The Activity Affective Complexity score is demonstrated in a subsample of older adults from the Health and Retirement Survey.
Extending our approach of choosing the next query in decision making studies: Testing probability weighting functions
Probability weighting functions relate objective probabilities and their subjective weights, and play a central role in modeling choices under risk within cumulative prospect theory. While several different parametric forms have been proposed, their qualitative similarities make it challenging to…
