Foundations of Data Science

Analysis, Statistical Modelling and Prediction

On this page: 

       Course description
       Target audience
       Course objectives
       Prerequisites
       Course content
       Preparation for the course
       References
       Instructor’s Bio

Duration: 12 hours/3 days 

An hour-long lunch break from 13:00-14:00 with RECSM Winter School participants and members

Format

Hybrid (in-person and online)

Lecturer

Burak Sonmez

Course Description

This intensive three-day course introduces participants to the foundations of social data science, combining statistical reasoning, computational tools, predictive modelling, and reproducible research practices for the analysis of social and behavioural data. The course is designed to help participants move from substantive or applied research questions to appropriate quantitative analyses and to interpret statistical results critically. It is relevant both to academic researchers and to researchers and analysts working in industry, government, NGOs, policy organisations, and other data-rich environments. Using R, participants will work with large real-world datasets. They will gain hands-on experience in exploring and visualising data, estimating and interpreting statistical models, investigating heterogeneity, generating predictions, and communicating results. The course introduces descriptive statistics and visualisation, linear regression, interaction effects, logistic regression, and an introductory application of random forests. These methods are situated within broader discussions of measurement, sampling, statistical uncertainty, prediction, explanation, and interpretation. Participants will also be introduced to reproducible research practices using R, RStudio, Git, and GitHub, enabling them to organise, document, version, and reproduce their analytical work. A central theme of the course is the distinction between description, explanation, and prediction. Participants will learn how different analytical approaches answer different types of questions and how the structure and quality of available data constrain the conclusions that can be drawn. Teaching combines lectures, demonstrations, guided coding exercises, and independent practical work. Participants will have opportunities to discuss their own research or analytical problems and consider how the methods covered could be applied in academic, policy, or industry settings.

Target Audience

The course is aimed at researchers, analysts, and postgraduate students who want to develop a practical foundation in social data science. It will be particularly relevant to participants from sociology, political science, social policy, education, public health, economics, psychology, international development, and related social and behavioural sciences.

The course is also designed for researchers and analysts working outside academia, including those in technology companies, market and consumer research, consulting, government, NGOs, international organisations, policy organisations, and other sectors where quantitative data are used to understand individuals, populations, organisations, or behaviour.

The course will be particularly useful for:

• Doctoral and Master's students developing quantitative research skills.
• Academic researchers seeking to strengthen their statistical and computational toolkit.
• Industry researchers and data analysts who want a stronger grounding in statistical reasoning and social-data analysis.
• Policy and government researchers working with observational or survey data.
• Researchers preparing for more advanced training in computational social science, machine learning, causal inference, or related methods.

No particular disciplinary background is required.

Course objectives

The course aims to provide participants with a practical and conceptual foundation for analysing social and behavioural data using modern quantitative and reproducible research practices.

By the end of the course participants will be able to:

• Translate substantive and applied research questions into quantitative analyses.
• Understand observations, variables, measurement, samples, and populations.
• Import, inspect, prepare, summarise, and visualise data using R.
• Work with both structured social surveys and larger, less curated real-world datasets.
• Use R, RStudio, Git, and GitHub as part of a reproducible analytical workflow.
• Understand distributions, statistical variation, and uncertainty.
• Estimate and interpret linear and logistic regression models.
• Work with continuous and categorical explanatory variables.
• Investigate heterogeneous relationships using interaction effects.
• Understand how flexible predictive approaches such as random forests can capture nonlinearities and complex interactions.
• Distinguish between explanatory and predictive modelling.
• Interpret predicted probabilities and model predictions.
• Understand the importance of training and evaluating predictive models appropriately.
• Recognise the distinction between association and causal inference.
• Understand how comparative, clustered, and survey data affect statistical analysis.
• Recognise the role of survey weights and representativeness.
• Present analytical findings clearly and responsibly.
• Develop a reproducible analytical workflow relevant to their own research or professional context.

Prerequisites

The course assumes general familiarity with quantitative research but does not require statistical or programming experience. Basic familiarity with R is strongly recommended but is not a formal prerequisite. Participants who have previously used R to load a dataset, inspect variables, or run simple commands will find it easier to concentrate on the substantive and methodological material. Participants without prior R experience are welcome to attend but will be encouraged to complete a short introductory R tutorial before the course. Pre-course materials will be provided or recommended for this purpose.
No previous knowledge of Git, GitHub, regression modelling, or machine learning is required. Familiarity with basic concepts such as variables, averages, percentages, and empirical research questions will be helpful.

Course Content

Preparation for the course

Participants should bring a personal laptop on which they have permission to install and run software.
Before attending, participants should install:

• A recent version of R.
• The current free version of RStudio Desktop.

No paid software or licences are required.
Participants with little or no previous experience of R will be provided with or directed to a short introductory tutorial and are strongly encouraged to complete it before the course.
Course datasets and the required R packages will be provided or specified in advance. The course will use European Social Survey data alongside a larger applied dataset to expose participants to different types of real-world analytical problems. 

References

There are no compulsory readings before the course. The course is designed to be self-contained, and participants are not expected to complete any of the readings in advance.
The following resources are recommended for participants who would like additional background or wish to explore particular topics in greater depth:

• Gelman, A., Hill, J., & Vehtari, A. Regression and Other Stories. Cambridge University Press.
  Recommended for regression modelling, interpretation, prediction, and thinking critically about what statistical models can tell us.
• Wickham, H., Çetinkaya-Rundel, M., & Grolemund, G. (2023). R for Data Science (2nd ed.). O'Reilly.
  Recommended for working with data in R, data visualisation, data transformation, reproducible workflows, and communicating analyses. The book is freely available online.
• James, G., Witten, D., Hastie, T., & Tibshirani, R. An Introduction to Statistical Learning: with Applications in R. Springer.
  Recommended for participants interested in prediction, statistical learning, model evaluation, and methods such as decision trees and random forests.
• Healy, K. Data Visualization: A Practical Introduction. Princeton University Press.
  Recommended for participants wishing to develop their skills in exploratory and explanatory data visualisation using R.

Instructor’s Bio

​

Burak Sonmez is Associate Professor in Quantitative Social Science at the Social Research Institute, University College London. His research examines how social and institutional contexts shape individual beliefs, identities, preferences, trust, and behaviour, and how these processes contribute to broader social outcomes. He uses experimental, quasi-experimental, and computational methods to study causal relationships in social science.
Before joining UCL, he was an LSE Fellow in Quantitative Social Research Methods at the London School of Economics. He holds a PhD in Sociology and an MA in Longitudinal Social Research from the University of Essex, and a Bachelor's degree in Economics from Bilkent University. His research has been funded by organisations including the British Academy, ESRC, Nuffield Foundation, and Education Endowment Foundation.