Spring 2024
The Quantitative Psychology colloquium series meets weekly in the Autumn and Spring semesters. The primary activity is presentation of ongoing research by students and faculty of the quantitative psychology program. Often, guest speakers from allied disciplines (e.g., Education, Statistics, and Linguistics) within and external to The Ohio State University present on contemporary quantitative methodologies. Additionally, discussions of quantitative issues in research and recently published articles are often conducted during colloquium meetings.
Faculty coordinator: Dr. Gyeongcheol Cho
Venue: Psychology 35 and online
Time: 12:30-1:30pm
Please contact Dr. Gyeongcheol Cho if you would like information on how to attend these events over zoom.
January 8, 2024
Organizational Meeting
January 15, 2024
Martin Luther King Jr. Day
January 22, 2024
Speaker: Dr. Marco Chen
Title: Modeling Construct Change Over Time Amidst Potential Changes in Construct Measurement: A Longitudinal Moderated Factor Analysis Approach
Abstract: A common interest in educational and psychological measurement is to examine change over time. To accurately capture true change in an underlying construct of interest, we must also account for changes in the way the construct manifests itself over time. In analyzing longitudinal data, however, growth curve models are often fit to a repeated measure constructed as a sum or mean of scale items, making an implicit assumption of constancy of measurement. This practice risks confounding actual construct change with changes in measurement (i.e., differential item functioning; DIF), threatening the validity of conclusions. An improved method that avoids such confounding is the second-order growth curve model (SGC), but the applicability of SGC is hindered by key limitations such as a lack of measurement model parsimony and difficulty in evaluating DIF from continuous covariates. Drawing on the advantages of moderated nonlinear factor analysis (MNLFA), we propose an alternative approach that provides a parsimonious framework for including many timepoints and DIF from different types of covariates. We implement this model through Bayesian estimation, allowing for incorporation of regularizing priors to facilitate efficient evaluation of DIF. We evaluate the performance of the proposed regularized longitudinal MNLFA model relative to growth models with mean scores through a simulation. We also demonstrate a workflow of measurement evaluation and growth modeling with an empirical example examining changes in adolescent delinquency over time.
January 29, 2024
Speaker: Dr. Ellen Hamaker
*joint event with University of Maryland, College Park; University of North Carolina at Chapel Hill, University of Notre Dame, Vanderbilt University; University of South Carolina; and University of Virginia
Title: Reflections on the Within-Between Dispute in Cross-Lagged Panel Research
Abstract: Cross-lagged panel data, involving the observation of two or more variables at multiple points in time within the same cases (e.g., individuals or dyads), provide unique opportunities to examine temporal stability and change. Recently, there has been a growing interest in various statistical models that dissect the observed variability in such data, distinguishing between stable between-person characteristics and temporal within-person fluctuations. While some researchers in the field of psychology enthusiastically endorse this approach as the right path forward, others have been more hesitant or sometimes even dismissive of these developments. In this presentation, I will examine the within-between distinction in panel research from three different angles: design, data, and research question. This exploration will underscore the importance of considering the timescale at which a process unfolds and how this relates to the study design and the stability patterns in the data.
February 5, 2024
Speaker: Dr. Heungsun Hwang
Title: An approach to structural equation modeling with both factors and components
Abstract: As psychology and many other sciences become interdisciplinary, there is an ever-increasing need to accommodate common factors and components in the same model and examine their relationships to understand human behaviour and cognition from more diverse perspectives. For example, researchers have increasingly been interested in the influences of genetic variation and/or altered brain activities on the variation of psychological constructs in cognition, personality, or mental disorders. Such psychological constructs have typically been considered common factors, whereas genetic or imaging constructs, such as genes and brain regions, have been considered components. However, existing methods for structural equation modeling (SEM) are not suitable for estimating models with both factors and components. Thus, my colleagues and I recently proposed a general SEM method, termed integrated generalized structured component analysis (IGSCA), which can also estimate such models. I will discuss IGSCA’s conceptual background and technical underpinnings and demonstrate its potential in real data applications with an investigation of the effects of multiple genes on depression severity. Moreover, I will briefly show how to apply the method using free, user-friendly software–GSCA Pro.
February 12, 2024
Speaker: Chia-Hsin Yin [Student Presentation]
Title: Vocabulary Size and Working Memory Effects on Bilingualism: Figurative Language Processing under fMRI.
Abstract: This study investigated the role of working memory capacity (WMC) in metaphoric and metonymic processing in Mandarin–English bilinguals’ minds. It also explored the neural correlations between metaphor and metonymy computations. We adopted an event-related functional magnetic resonance imaging (fMRI) design, which consisted of 21 English dialogic sets of stimuli and 5 conditions: systematic literal, circumstantial literal, metaphor, systematic metonymy, and circumstantial metonymy, all contextualized in daily conversations. Similar fronto-temporal networks were found for the figurative language processing patterns: the superior temporal gyrus (STG) for metaphorical comprehension, and the inferior parietal junction (IPJ) for metonymic processing. Consistent brain regions have been identified in previous studies in the homologue right hemisphere of better WMC bilinguals. The degree to which bilateral strategies that bilinguals with better WMC or larger vocabulary size resort to is differently modulated by subtypes of metonymies. In particular, when processing circumstantial metonymy, the cuneus (where putamen is contained) is activated as higher-span bilinguals filter out irrelevant information, resorting to inhibitory control use. Cingulate gyrus activation has also been revealed in better WMC bilinguals, reflecting their mental flexibility to adopt the subjective perspective of critical figurative items with self-control. It is hoped that this research provides a better understanding of Mandarin–English bilinguals’ English metaphoric and metonymic processing in Taiwan.
February 19, 2024
Speaker: Dr. Mehta Paras
Title: Structural Equation Modeling of Relational Data-Structures
Abstract: Social Relations Model (SRM) involves reciprocal dyadic ratings where each individual rates every other individual in a small group. Such data challenge our notions of validity implicit in conventional self-reported measures of personality, psychopathology and employee ratings data. Methodologically, reciprocal ratings data involve complex dependencies that cannot be modeled within current Multilevel Modeling (MLM) or Structural Equations Modeling (SEM) frameworks. A broader conceptual, methodological and computational framework is necessary to specify and estimate SRMs. This presentation will gently introduce the notion of SEM models of Relational Data Structures or n-Level Structural Equation Modeling. NL-SEM includes novel methodological constructs such as ‘virtual levels’ and ‘role models’ that allow a natural specification of latent variable SRM models as a six-level model. An empirical application of an explanatory latent variable SRM model of agreeableness will be presented. The following diagram illustrates a latent variable SRM with four real levels and two virtual levels. A survey of applications of NL-SEM to other areas of psychology will be presented.
NL-SEM is applicable to a wide range of complex multilevel data-structures (e.g., crossclassified, switching membership etc.). NL-SEM framework is implemented in an R-package called xxM (http://xxm.times.uh.edu). xxM provides an intuitive ‘graphical’ and ‘interactive’ user-interface that makes the task of specifying complex models easy.
February 24, 2024
Speaker: Dr. Emilio Ferrer
*joint event with University of Maryland, College Park; University of North Carolina at Chapel Hill, University of Notre Dame, Vanderbilt University; University of South Carolina; and University of Virginia
Title: Intra- and Inter-Individual Variability in Psychological Processes in Dyadic Interactions
Abstract: The study of intra-individual variability is well recognized as a crucial premise to understanding individual processes. On the strength of this principle, psychologists have devoted substantial effort to examining and comprehending the essence unique to the person (e.g., Allport, 1937). Evidence of the pursuit of this aim are the methods developed to capture an individual’s fluctuations over time. In this talk, I describe several of such methods and extend them to the study of dyadic interactions. I focus on how these methods can help to identify patterns of intra- and inter-individual variability in dyads and how this information can be used to make predictions about future states of the system. I conclude with current challenges and ideas for future research.
March 4, 2024
Speaker: Dr. Jiawei Xiong
Title: Next-Generation Process Data Innovations with the introduction of Sequential Reservoir Model
Abstract: This talk centers on the analysis of assessment process data. There has been a growing interest in utilizing process data from log files in computerized assessments. This type of data, which captures examinees’ response activities such as clickstreams during tests, provides insights into their engagement and problem-solving trajectories. This talk introduces a recent advanced model, Sequential Reservoir Model, based on data-driven approaches to extract interpretable features from unstructured process data. These features can clearly distinguish examinees’ behavior patterns. This innovative analysis of process data can help assess latent variable analysis in measurement, thereby providing a more comprehensive measurement of examinee performance.
March 11, 2024
Spring Break
March 18, 2024
Speaker: Shannon Jacoby [Student Presentation]
Title: Assessing Monotonicity and Trends: A Simulation Study
Abstract: Across a wide array of substantive research areas within psychology, many researchers make use of the regression modeling framework. A central assumption of this framework is that the relationship between the IV and expected value of the DV is linear, and the current diagnostic criterion for this assumption is a visual inspection of a scatter- or residual plot for the absence of systematic nonlinearity. Given that the violation of the linearity assumption can lead to biased parameter estimates and error variances, the current work endeavors to create an additional avenue for examination of the linearity assumption that goes beyond graphical analysis. Borrowing inspiration from the ANOVA framework, contrast analyses were applied to simulated data as a way to detect linear and quadratic trends in continuous data as well as nonmonotonicity via an ordinal analysis approach. As not all monotonic relations are linear and not all quadratic trends are violations of monotonicity, this work investigates both trends and violations of monotonicity in an effort to more fully understand IV/DV relationships under a variety of conditions. Type I errors, power, and comparison of findings with additional methods are discussed, as well as limitations and directions for future research.
March 25, 2024
Speaker: Dr. Yves Rosseel
*joint event with University of Maryland, College Park; University of North Carolina at Chapel Hill, University of Notre Dame, Vanderbilt University; University of South Carolina; and University of Virginia
Title: The structural-after-measurement (SAM) approach to SEM
Abstract: In structural equation modeling (SEM), the measurement and structural parts ofthe model are usually estimated simultaneously. But already since the birth of SEM in the ’70s, various authors have advocated that we should first estimate the measurement part, and then estimate the structural part. We call this the Structural-After-Measurement (SAM) approach. In the first part of the presentation, I will give a brief historical overview of various SAM approaches, and discuss their advantages and disadvantages. Next, I will describe the so-called `local’ SAM method where the mean vector and variance–covariance matrix of the latent variables are expressed as a function of the observed summary statistics and the parameters of the measurement model. The method includes two-step corrected standard errors and local fit measures. In the second part of the presentation, I will discuss several recent developments that are based on the SAM approach, including the inclusion of latent quadratic and interaction terms, the use of non-iterative estimators for the measurement part of the model, small-sample corrections, and various approaches to study measurement invariance in the setting where the number of groups is very large. Finally, I will discuss a software implementation of the SAM approach that is available in the R package lavaan.
April 1, 2024
Speaker: Dr. Christopher Urban
Title: Progress and Problems in Deep Learning-Based Latent Variable Modeling
Description: Over the past few years, researchers have made progress in understanding how deep learning methods can be used to enhance traditional latent variable models by making these models more flexible and faster to fit. However, many interesting problems have not yet been thoroughly investigated. In this talk, I’ll discuss the recent progress and demonstrate new software for fitting deep learning-based latent variable models. I’ll conclude with some preliminary work on two problems I am investigating for my dissertation: modeling (1) non-normal latent variables and (2) intensive longitudinal data.
Title: A Deep Learning Approach to Modeling Intensively Measured, Longitudinal, and Multidimensional Item Responses
Description: Smartphones and wearable sensors now afford social scientists unprecedented opportunities to collect intensive longitudinal data (ILD). ILD consisting of participant responses to repeated assessments taken at short intervals is useful for gaining insights into the fine-grained dynamics of experiential and behavioral processes. In this talk, I’ll discuss my recent work on developing a deep learning framework for second-order latent growth modeling using ILD. Specifically, my approach combines stochastic differential equations (SDEs; a broad class of continuous-time time series models) with neural networks to produce second-order latent growth models (SLGMs) with greatly enhanced modeling capacity relative to conventional methods. In addition to more flexible SLGMs, this new framework includes models for the temporal distribution of measurement occasions and response missingness as well as statistical tests for goodness-of-fit and model selection. I will showcase some recent simulation results as well as an application to modeling emotion-switching in individuals with borderline personality disorder.
April 8, 2024
Speaker: Frank Leyva Castro [Student Presentation]
Title: A Structural-After Measurement Approach: Extending Graphical Diagnostics to Structural Equation Modeling
Abstract: This presentation delves into the research on graphical diagnostics in Structural Equation Modeling (SEM). Using existent methods in other linear models, a strategic approach based on a Structural-After-Measurement is proposed to evaluate model assumptions: The measurement model is examined before evaluating the assumptions among latent variables. By using Thurstone factor score estimates for residual calculations, an analogous procedures used in Multilevel modeling is followed. An empirical example is used to illustrate the application of this approach highlighting both the results of this procedure and the challenges encountered.
April 15, 2024
Speaker: Dr. Lily Hu, Assistant Professor of Philosophy, Yale University.
*joint event with University of Maryland, College Park; University of North Carolina at Chapel Hill, University of Notre Dame, Vanderbilt University; University of South Carolina; and University of Virginia
Title: Equality, Identity, and Causality
Abstract: In this talk, I argue that whether something constitutes disparate treatment on the basis of race or sex (or other socially salient group classifications) cannot be spelled out purely causally. This is because whether something constitutes similar treatment or different treatment on the basis of race (or sex, etc.) is a non-trivial normative matter that is not made easier by our causal scalpels. It depends, rather, on how one characterizes the treatment or the principle that some treatment embodies. One must, first, characterize what the relevant kinds of similarity and difference are in order to determine whether two individuals who are differently racialized (or differently sexed, etc.) are indeed treated similarly or differently. That this step of defining the relevant notions of similarity and difference is an irreducibly normative one is a truism, but its implications have not been fully appreciated by contributors to causal inference work on fairness and discrimination.
April 22, 2023
Brief presentations on external talks by the following students:
Frank Leyva Castro
Shannon Jacoby
Chia-Hsin Yin