Advanced Statistical Methods And Models In Experimental Design 2500-EN-CS-SM-03
This course provides a comprehensive dive into the statistical tools essential for modern cognitive science research. Expanding upon the basic General Linear Model, the course moves into Generalized Linear Models (GLM) to handle categorical and non-normal data (e.g., accuracy, reaction time distributions).
A significant portion of the course is dedicated to Linear Mixed Effects Models (LMEM), which have become the standard for analyzing hierarchical and repeated-measures data typical in cognitive science experiments. Students will also be introduced to Structural Equation Modelling (SEM) to test complex theoretical structures and to basic meta-analytical techniques used for synthesizing results from multiple studies.
The course adopts a "hands-on" workshop philosophy. With the expanded 60-hour format, classes are split between theoretical introductions and extensive individual and group coding sessions. Students will work with real datasets, focusing not just on running tests, but on data simulation, model diagnostics, visualization, and the principles of reproducible research using R (Tidyverse ecosystem) and RMarkdown/Quarto.
Learning activities:
1. Interactive Lectures: Introduction of statistical concepts using slides and live coding demonstrations.
2. Hands-on Workshops: Students solve problem sets on their own computers during class with immediate instructor feedback.
3. Live Coding: Real-time problem solving and debugging demonstrations.
4. Data Simulation: writing R scripts to simulate experiments to understand power and model behavior.
5. Project Work: Working on real datasets to apply the full analytical pipeline.
Learning outcomes
After successfully completing the course, students will be able to:
1. Model selection & justification
Select and justify appropriate statistical models for cognitive-science experimental designs (e.g., regression/ANOVA framework, GLMs for non-normal outcomes, LMEM/GLMM for repeated measures, SEM/CFA), explicitly stating key assumptions and limitations. (K_W03, K_W04, K_U03)
2. Implementation in R (end-to-end workflow)
Implement a complete analysis workflow in R (import/tidy, visualize, fit models, extract results) using modern tools (e.g., tidyverse, lme4, lavaan, Quarto/RMarkdown) in a way that supports reproducible output. (K_W07, K_U07, K_U12)
3. Diagnostics, robustness, and error detection
Diagnose and validate fitted models (assumption checks, residual diagnostics, influence/outlier analysis, convergence checks; simulation/bootstrapping when appropriate) and revise analytical choices accordingly. (K_W03, K_U03, K_U07, K_K01)
4. Interpretation & scientific reporting
Interpret model parameters and uncertainty (e.g., coefficients/interactions, random effects, odds ratios, effect sizes and CIs) and communicate results clearly in a structured scientific report with appropriate tables/figures. (K_U07, K_U11, K_U12)
5. Reproducible & responsible research practice
Organize code, data, and reporting to enable verification and reuse (transparent analytic decisions, documented workflow, versioned/reproducible outputs), adhering to professional standards of research work. (K_W07, K_U12, K_K06, K_K07)
Assessment criteria
a) Assessment methods:
- Homework Assignments: A series of practical coding assignments (e.g., 3-4 assignments) distributed throughout the semester. These require students to analyze provided datasets, write clean R code, and interpret the results.
- Final Project: An independent analysis of a complex dataset (provided by instructor or student's own). Includes a fully reproducible script (RMarkdown/Quarto) and a written report interpreting the findings.
- Active Participation: Constructive engagement during workshop sessions.
b) Components of the final grade and their weights:
- Homework Assignments: 50%
- Final Project: 40%
- Active Participation: 10%
c) Grading scale:
- 0-50%: 2 (Fail)
- 51-60%: 3 (Satisfactory)
- 61-70%: 3.5 (Satisfactory+)
- 71-80%: 4 (Good)
- 81-90%: 4.5 (Good+)
- 91-100%: 5 (Very Good)
d) Requirements for retaking the assessment:
- Homeworks: Late submissions may be penalized. Failed assignments may be corrected and resubmitted within a specified timeframe (e.g., 2 weeks) for a capped grade.
- Final Project: If the project is failed, the student must submit a revised version or a new analysis within the retake session period.
e) Exams in the exam session:
i) Requirements for taking the exam: Not applicable. This course uses continuous assessment (assignments + project). There is no separate sit-down exam. Completion of all homework assignments is required to submit the final project.
ii) Possibility for retaking the exam in case of a positive grade: Not applicable (no exam).
iii) Early Exam Session ("Zerówka"): Not applicable. The course is project-based. The "Zero" session may be treated as an early deadline for the Final Project submission if agreed upon with the instructor.
Attendance ruels:
Attendance is mandatory for workshop sessions. Students are allowed a maximum of 2 absences without a formal excuse and additional 2 with a formal excuse. Any missing in-class activity requires completing makeup assignments or may result in failing the course.
Additional information
Additional information (registration calendar, class conductors, localization and schedules of classes), might be available in the USOSweb system: