PSICOSTAT 3.1

Meetings' calendar

29th May 2020

Ottavia Epifania: Shine bright like an open source app: An introduction to shiny

R (R Core Team 2018) provides an Open source alternative for performing advanced data analysis in a straightforward and accessible way. However, R might be not straightforward for people with a background in Social Sciences, who might find the basic programming skills required for its use overwhelming and discouraging. Fortunately, shiny package (Chang et al. 2020) allows for creating interactive and easy to sue web application while exploiting R computational power. In this way, it is possible to provide open source and accessible web applications without requiring any programming skills from the users. In this brief tutorial, the basics concepts and functioning of shiny are presented. You can find the code I used for the tutorial and teh code for an example app on my GitHub profile.

Please, click here for the codes of the presentation.

Want to know more? Contact Ottavia at otta.epifania@gmail.com

Giulia Calignano: Get up, stand up for your registered reports: an original study on the impact of audiovisual correspondences on the attentional system in 12-month-old infants.

The pre-registration challenge was set by scientific researchers in 2017 - 2018 with the goal to stem the devastating effects of the publish or perish motto. The pre-registration process challenges researchers to submit to peer-review theoretical and methodological details of their researches prior to the actual data collection. Such Copernican’s revolution aimed at reducing the publication bias, the replication crisis and the bad practices that lead to less trustworthy science. That is, to allow the publication of sounding researches regardless of the outcomes. Here, I present my personal pre-registration challenge as a first years PhD student in cognitive sciences. I proposed a study aiming at investigating the impact of audiovisual correspondences, i.e. novel objects paired with unusual names, on the attentional system of 12-month-old infants. The measures of pupil size variation and saccade movements were selected as good indexes of resources allocation and disengagement of attention, respectively. The study aimed to move a first step into the understanding of the degree by which linguistic information held in working memory guide subsequent visual attention deployment in infants. In this talk, I share the dark and the bright side of my experience. The overall conclusion suggest that the pre-registration process did improve the theoretical and methodological reflection of the research inquiry. Nevertheless, we have a long way to go to make the pre-registration process sounding for young researchers in psychology and cognitive sciences.

Please, click here for the presentation slides.

Want to know more? Contact Giulia at calignano.giulia@gmail.com

13th May 2020

Margherita Andrao: Dancing Numbers: Exploring dancers' space-number associations

Recently, some studies have highlighted the impact of body experience in numerical cognition (embodied numerosity) and the possible relation among numerical, spatial, and temporal dimensions (ATOM). Based on these findings, the general purpose of this project is to explore whether the sensory-motor training in dance could modulate the numerical representation. In particular, our main interest is to study how the spatial-numerical association of response codes (SNARC effect) differs between expert dancers and non-dancers. Our second aim is to explore how different dance styles (ballet and freestyle hip hop), characterized by different counting habits of music and movement, can influence this numerical representation. We will recruit three groups of volunteers: 30 expert ballet dancers (experimental group), 30 expert freestyle hip hop dancers (control group with sensory-motor training), and 30 non-dancers (control group). All volunteers will be between 18-35 years old, female, right-handed, and without histories of neuropsychological disorders or substance abuse. Participants in each group will perform the online version of the experiment consisting of a parity judgment task using numbers 1 to 8 (experimental task), and of the Simon effect task (control task).

Please, click here for the presentation slides.

Want to know more? Contact Margherita at margherita.andrao@studenti.unipd.it

Alessio Toraldo: Measuring neglect and realizing that many times it's not there

Unilateral neglect is the inability to properly process, or respond to, stimuli that are presented on one side of space after a lesion of the opposite hemisphere. In decades of research a huge number of tasks have been proposed, which provide the same data structure: a set of stimuli is presented, each of which is associated with a spatial position and a Hit/Miss score. Despite this homogeneity, no common statistical procedure for measuring and diagnosing neglect on these tasks has been proposed. The aim of this work is to provide some uniformity. In this lecture I show: (i) that the Mean Position of Hits (MPH) is an optimal measure of neglect for several theoretical reasons; (ii) that (as all measures) MPH is expected to produce an inflation of false positive rates when deficits other than neglect are present, and a classical diagnostic procedure relying on a control sample is used; (iii) empirical evidence of such inflation; (iv) a statistical solution to this problem, based on a large simulation study. These efforts yielded an equation that provides precise estimation of the expected within-patient instability of MPH given the severity of non-neglect deficits. Taking into account such instability in diagnosis allows one to bring false positive rates back to nominal values (e.g. 5%). Under some assumptions, which proved to be empirically close to correct, this new diagnostic method does not need a control sample, but only the data from a single patient. This data can be inserted in a worksheet (http://psicologia.unipv.it/toraldo/mean-position-of-hits.htm) which directly gives neglect diagnosis as an output.

Please, click here for the presentation slides.

Want to know more? Contact Alessio at alessio.toraldo@unipv.it

30th April 2020

Irene Valori: Exploratory approach to the study of typical development and Autism Spectrum Disorders

When learning and interacting with the world, we integrate the sensorimotor information coming from both the external environment and our body. The reliability of this information varies across developmental ages and trajectories, with early atypical profiles in the case of neurodevelopmental disorders such as Autism Spectrum Disorders. This talk presents the potential use of Immersive Virtual Reality and head-mounted displays, which have unique characteristics that enable the manipulation of sensorimotor features of the environment, according to research and clinical purposes. We will present the advantages of an exploratory approach to this research topic, the theoretical and practical sides of it. We will see two exploratory studies that provide first insights on how children and adults with typical development or Autism differently interact with real and immersive virtual environments. We will talk about feasibility issues, descriptive statistics, graphical inspection of data, group representativeness versus meaningful individual variability.

Please, click here for the presentation slides.

Want to know more? Contact Irene at irene.valori@hotmail.it

Enrico Toffalini: The "incredible cure" for dyslexia: too good to be true, perfect for publishing (...and how to move on)

Assessing treatment efficacy in dyslexia is challenging. The effects are (almost by definition) small, but large samples are difficult to recruit. Nearly all existing literature is underpowered, leading to overestimated effect sizes and unwarranted claims frequently based on convenient multiple testing. Using Bayes Factor instead of p-value as the inferential criterion does not solve the problem. Rather, an appropriate use of mixed-effects models (instead of traditional ANOVAs) can be largely beneficial. Moving the focus from the group to the individual level is potentially the key. Repeated measures can be strategically used at different time points to estimate precise individual parameters and treatment gains even with small samples. Examples of design analysis conducted via simulation are shown.

Want to know more? Contact Enrico at enrico.toffalini@yahoo.it

15th April 2020

COVID-19 Talks

Gianmarco Altoè: COVID-19: Data Talk! But what are they saying? 7 hints

Please, click here for the presentation slides. Video recordings will be available soon!

Paolo Girardi: COVID-19: An epidemiological perspective

Please, click here for the presentation slides. Video recordings will be available soon!

16th December 2019

Luca Menghini: Towards multilevel CFA: design analysis and cross-level invariance.

The experience sampling method (ESM) is the repeated sampling of current psychological states, daily activities, and interactions to provide information on their frequency, intensity, and patterning. ESM is increasingly used, for instance, in workplace stress research to overcome some limitations of retrospective self-report measures. However, little guidance has been provided on the evaluation of ESM scales' psychometric proprieties, and most studies use mono-item or shortened versions of retrospective scales without taking into account the multilevel data structure. With the aim to provide a step-by-step procedure on the validation of ESM scales, the presentation focuses on the multilevel confirmatory factor analysis (MCFA). The basics of MCFA are described, with a focus on the between- and within-cluster covariance matrices computation, the design analysis, and the cross-level invariance evaluation.

Want to know more? Contact Luca at luca.menghini.3@studenti.unipd.it

Angela Andreella: First steps with Python.

We will see the basic concept of Python, which type of IDE and GUI choose, how Anaconda works, and how to create an environment and install the packages. Then, we will explore the essential function of the packages numpy, pandas, matplotlib, seabron, scikit learn, and statmodels using some datasets downloaded from the kaggle webpage. We will perform a first simple exploratory analysis, then a linear regression analysis, principal components analysis, and classification analysis. You can find the script executed using Jupyter notebook in my Github account.

Want to know more? Contact Angela at angela.andreella@phd.unipd.it

Simone D'Ambrogio: Statistical analysis to study how attention affects decision-making processes.

2nd December 2019

Massimiliano Pastore: The power of power

Statistical power is the probability of achieving the goal of an empirical study, when a suspected underlying state of the world is true. Unfortunately, when there are multiple or multivariate hypotheses, or when models are complex with many parameters, the definition and the computation of power is not straightforward. In this presentation I will show, with two different examples, how to estimate the power for detecting the target effects via simulation.

Want to know more? Contact Massimiliano Pastore at massimiliano.pastore@unipd.it

Paolo Girardi: Functional data analysis of eye pupil dilation

In general, methods for correlating pupillary response to the cognitive activity of a subject undergoing an evaluation of cognitive activity are based on average values or analysis of peak activity. Eye tracking data contains several features: Eye point of gaze; Fixation (and relative Area Of Interest -AOI-) and Saccade time periods; Eye pupil dilation. Differently by the classical approach based on differences on “averaged” values over a certain period,we propose a functional data clustering approach that takes into account the entire behaviour of a eye dilation time series.

You can find out more here. Contact: paolo.girardi@unipd.it

It started

18th November 2019

Psicostat is back!

Gianmarco Altoè: Psicostat 3.1: Welcome Presentation

The Psicostat group is back! We look forward to a year of discussions and reciprocal exchange between the fields of Psychology and Statistics, always keeping in mind our motto: "Accept uncertainty, be Thoughtful, Open and Modest. Remember: ATOM

Want to know more? Look at our "Meetings' calendar".

Simone D'Ambrogio: Creating BeardPlot in R using ggplot2

The gg_beardplot function provide ad easy tool based on ggplo2 to visualize the distribution of the data and its probability density. It is similar to the violin-plot, the rain-cloud-plot, the bean-plot among others, but the gg_beardplot is inspired by the famous Fisher beard ;)

Want to know more? Look at Simone's GitHub repository here. Contact: simone.dambrogio@studenti.unipd.it

Filippo Gambarota: Making a peRsonal website with Hugo and Github

It is possible to build personal and academic website using Hugo and the Academic theme! Introduction slides and some links and resources to get started with Hugo are available here. This is not a complete guide to Hugo or a deep tutorial about managing a static website but simply a starting point.

Want to know more? Look at Filippo's GitHub repository available here. Contact: filippo.gambarota@phd.unipd.it

Livio Finos: Contrasting Contrasts

I stressed the importance of using zero-sum contrasts when dealing with categorical variables (e.g. factors in an experimental design) compared to the usual dummy coding. Analogous issues are present when considering quantitative variables. I illustrated this issue with a simple dataset and a linear model including one factor (i.e., categorical variable), one quantitative variable and their interaction.

Want to know more? Look at the presentation available here.