Artificial Neural Networks for Short-Form Development of Psychometric Tests: A Study on Synthetic Populations Using Autoencoders

Casella, Monica; Dolce, Pasquale; Ponticorvo, Michela; Milano, Nicola; Marocco, Davide

doi:10.1177/00131644231164363

Cited by 4 publications

(2 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Despite the growing popularity of predictive techniques in psychology and social sciences (e.g., Casella et al, 2024;Dolce et al, 2020), statistical modelling for explanation remains the predominant approach. Conversely, in domains such as bioinformatics and natural language processing, algorithmic modelling predominates Yarkoni & Westfall, 2017).…”

Section: Explanatory Modelling and Predictive Modellingmentioning

confidence: 99%

Exploring predictive clustering: A multi-step approach with finite mixture models and machine learning

Robba,

Casella,

Iannello

et al. 2024

Preprint

View full text Add to dashboard Cite

Finite mixture models are clustering methods gaining more and more popularity recently. They also show many advantages in comparison to traditional clustering techniques (e.g., k-means cluster analysis). However, contrarily to techniques relying on classification algorithms, mixture models are not able to classify unseen or future cases in clusters previously identified. Hence, this study proposes a novel multistep approach to predict group membership using clusters obtained through a Latent Profile Analysis (LPA) as reference labels. A representative sample of 1,002 Italian individuals was used, with 802 participants randomly selected for initial analysis. Four Machine Learning (ML) algorithms - Artificial Neural Networks, Random Forest, Gradient Boosting, and Support Vector Machines - were trained and evaluated, relying on grid search and k-fold cross-validation. Results indicated high accuracy across algorithms, with Support Vector Machines (SVM) exhibiting the highest performance (95.3%). To further assess generalization capacity, a second subset of 200 individuals was analysed, with labels suggested by a second LPA from the total dataset (N = 1,002) used as reference. Trained ML models achieved notable accuracy in predicting cluster membership for this second subset (SVM accuracy = 90%). Overall, the study demonstrates ML efficacy in predictive clustering, offering a robust framework for future research.

show abstract

Section: Explanatory Modelling and Predictive Modellingmentioning

confidence: 99%

Exploring predictive clustering: A multi-step approach with finite mixture models and machine learning

Robba,

Casella,

Iannello

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…In particular, artificial neural networks have been used to select variables for inclusion in a psychopathological model [42] and for the development of short forms of tests [43]. Staying within the methodological realm, the demonstrated ability of autoencoders to extract essential information from data has paved the way for new applications of autoencoders.…”

Section: Artificial Neural Network In Psychometricsmentioning

confidence: 99%

Exploring the Potential of Variational Autoencoders for Modeling Nonlinear Relationships in Psychological Data

Milano,

Casella,

Esposito

et al. 2024

Behavioral Sciences

View full text Add to dashboard Cite

Latent variables analysis is an important part of psychometric research. In this context, factor analysis and other related techniques have been widely applied for the investigation of the internal structure of psychometric tests. However, these methods perform a linear dimensionality reduction under a series of assumptions that could not always be verified in psychological data. Predictive techniques, such as artificial neural networks, could complement and improve the exploration of latent space, overcoming the limits of traditional methods. In this study, we explore the latent space generated by a particular artificial neural network: the variational autoencoder. This autoencoder could perform a nonlinear dimensionality reduction and encourage the latent features to follow a predefined distribution (usually a normal distribution) by learning the most important relationships hidden in data. In this study, we investigate the capacity of autoencoders to model item–factor relationships in simulated data, which encompasses linear and nonlinear associations. We also extend our investigation to a real dataset. Results on simulated data show that the variational autoencoder performs similarly to factor analysis when the relationships among observed and latent variables are linear, and it is able to reproduce the factor scores. Moreover, results on nonlinear data show that, differently than factor analysis, it can also learn to reproduce nonlinear relationships among observed variables and factors. The factor score estimates are also more accurate with respect to factor analysis. The real case results confirm the potential of the autoencoder in reducing dimensionality with mild assumptions on input data and in recognizing the function that links observed and latent variables.

show abstract

Autoencoders as a Tool to Detect Nonlinear Relationships in Latent Variables Models

Esposito,

Casella,

Milano

et al. 2023

2023 IEEE International Conference on Metrology for eXtended Reality, Artificial Intelligence and Neural Engineering (MetroXRAI

View full text Add to dashboard Cite

Artificial Neural Networks for Short-Form Development of Psychometric Tests: A Study on Synthetic Populations Using Autoencoders

Cited by 4 publications

References 61 publications

Exploring predictive clustering: A multi-step approach with finite mixture models and machine learning

Exploring predictive clustering: A multi-step approach with finite mixture models and machine learning

Exploring the Potential of Variational Autoencoders for Modeling Nonlinear Relationships in Psychological Data

Autoencoders as a Tool to Detect Nonlinear Relationships in Latent Variables Models

Contact Info

Product

Resources

About