Workshops on data analysis

The meetings are organized by Enrico Toffalini in collaboration with the PsicoStat group. Some additional materials can be found on github.

21st of February, 2024

Interactions and effect sizes: how to interpret them?

Why is this important? In the last meeting we provided some fundamental insights into interpreting effect sizes for “main effects.” However, a large portion of the research we conduct is not aimed at “discovering” main effects, which are often already well known, but at “discovering” the factors that moderate those effects, namely, to “discover” interactions. But how should effect sizes be interpreted in interactions?

Interaction and effect sizes: click here to see the Rmarkdown

Click here to see an additional example with power simulation.

Fully open code on GitHub here

2nd of February, 2024

Effect Sizes: interpretation and reporting from simple to complex

Cohen’s d and correlation: click here to see the Rmarkdown

Effect size in mixed-model and glm: click here to see the Rmarkdown

Some additional materials: click here

click here to download the full Rmarkdown code and materials

15th of November + 13th of December, 2023

Longitudinal models for panel data: R code, extensions, and examples

Click here to see a preview of the slides

see here a supporting Rmarkdown with simulation and model fitting

dataset for exercise: download (NOT the one in the slides)

(prior meeting slides are here)

30th of November, 2023

Power simulation: a real case

Click here to see the slides

see an supporting markdown here

31st of October, 2023

GLM, overdispersion, and the “flipscores” package

Click here to see the introductory Rmarkdown

19th of October, 2023

Longitudinal models for panel data (with few waves): What research questions can we address?

Click here to see the slides

also, enjoy this shiny app that supports the slides

13th of September, 2023

Assessing inferential risks when Clustering, and designing a Shiny App for a tutorial

Click here to see the slides

Click here to download the R code

Click here to see an additional markdown and here for a related article

06th of September, 2023

Power analysis for LMM: a real case of power simulation

Click here to see the slides

Click here to download the R code

28th of June, 2023

Livio Finos: Permutation testing… including repeated measures and covariates

Click here to see the slides

07th of June + 14th of June, 2023

Enrico Toffalini, Massimiliano Pastore: Basic tutorial on “brms”

Click here to see the slides

Click here to download the R code

17th of May, 2023

Enrico Toffalini and Filippo Gambarota: Tutorial on simulating mixed-effects models for designing multi-lab studies (and meta-analyses)

What is similar between a mixed-effects model, a multi-lab study, and a meta-analysis? How can we simulate these scenarios? Here is some slides and code:

Click here to see the slides

Click here to download the R code

19th of April, 2023

Umberto Granziol: Analyses with and without planned comparisons: some practical examples

A very important and applicative topic! Customizing contrasts in our linear models may help make it our hypothesis testing much more straightforward, and helps us interpret the model parameters. Practical examples are presented on simulated data.

Click here to download the datasets and code for the exercises

22nd of March + 5th of April, 2023

Gianmarco Altoè: Logistic regression (and generalized linear models in Psychology)

Very often, response variables in psychology are not generated by Normal distributions, which leads to fundamental violations of the assumptions needed for linear models. Luckily, GLMs (generalized linear models) are available for most situations in which responses are known to be non-normal. Among them, logistic regression is especially important as it allows us to model probabilities for binomial 0 / 1 data (e.g., the probability of observing correct responses as a functions of predictors in an experiment).

Click here to see the slides (22/03/2023)

(Also, see this very important series of slides by prof. Altoè on generalized linear mixed-effects models)

In the second meeting (05/04/2023), some researchers have presented their analyses using mixed-effects logistic regressions.

8th of March, 2023

Practical Exercises on Linear Mixed Models

Click here to see the slides (08/03/2023)

Click here to download the datasets for the exercises (08/03/2023)

22nd of February, 2023

Introduction to Linear Mixed Models

Linear mixed models (LMM) arguably represent the ideal framework for data analysis in most actual research scenarios in psychology. LMM are more flexible and they extract more information from data than common alternatives when they are used to their full potential. When their logic is understood, they are also a powerful aid in the phase of planning new research.

Click here to see the slides (22/02/2023)

Click here to download the R code (22/02/2023)

15th of February, 2023

The “for” loop, and how to run power analysis with simulated data for simple linear models

Once the basics facts about coefficients of linear models are understood (see first two meetings), it’s time to run a power analysis. The “for” loop is an intuitive tool for running power analysis using simulated data. In fact, power could be calculated analytically for simple cases, but understanding how to use simulation and the “for” loop opens the way to applying it to more complex scenarios as well. The cases we see here involve basic simple and multiple linear regression models.

Click here to download the slides and R code (15/02/2023)

18th of January + 1st of February, 2023

Coefficients in linear models, interactions, model comparisons, and data simulation

Understanding coefficients in linear models is the starting point for many more complex skills in data analysis. In the first two meetings of this workshop on data analysis, we discussed the interpretation of linear model coefficients in a series of cases featuring models with main effects, with interactions between two category variables and interactions between a continuous and a category variable. Also, we briefly discussed model comparisons and we started practicing some data simulation for the purpose of performing power analysis in the next meeting(s).

Click here to download the slides and R code of the first meeting (18/01/2023)

Click here to download the slides and R code of the second meeting (01/02/2023)