Oxford Spring School Course Sessions
The Oxford Spring School will run from Monday 19 March – Friday 23 March 2018.
The two main courses in the 2018 Oxford Spring School will be:
1) Casual Inference
2) Spatial Analysis
Further information will be made available shortly. We aim to open applications in early December 2017, with applications due in January 2018.
|Causal Inference||Causal Inference||Causal Inference||Causal Inference||Causal Inference|
|Spatial Analysis||Spatial Analysis||Spatial Analysis||Spatial Analysis||Spatial Analysis|
One Hour Walk-in
|'R' Walk-in clinic||'R' Walk-in clinic||'R' Walk-in clinic||'R' Walk-in clinic||'R' Walk-in clinic|
|Forecasting||Forecasting||Forecasting||Data Visualisation||Data Visualisation|
|Fieldwork Techniques||Fieldwork Techniques||Fieldwork Techniques||Process Tracing||Process Tracing|
(to be announced)
What does space add to our understanding of political phenomena? What can we learn from investigating how units (states, individuals, nodes of a network) interact with each other in space? The course is intended for Social Scientists who want to take advantage of spatial data to enrich their analyses by combining newly available georeferenced data or creating their own. Participants will also learn how to detect and model spatial dependencies that most social phenomena exhibit. The three main objectives of this course are i) building datasets based on geographical/spatial features of phenomenon of interests, ii) visualizing spatial patterns and map clustering using Moran’s Index and Local Indicators (LISA), and iii) modelling spatial interdependence to make more accurate statistical inferences. The first part of the course focuses on how to use GIS tools to build spatial dataset containing geographical variables (e.g. distances) and to combine (and potentially geocode) different types of data based on locations. The second part of the course moves to the statistical analysis of such data, with particular focus on the detection of spatial dependencies and their modelling using spatial lag and spatial error models.
Most social scientists learning statistics will come across the following statement: “Correlation does not imply causation”. Indeed, even the most sophisticated statistical models rarely identify causal effects. Randomized experiments that are the gold standard to do so are difficult—or even impossible—to implement. So, how can we understand and identify causal relationships? This course puts forward the idea of causal inference using potential outcomes and shows participants research designs to identify causal effects. The course will begin from simple experimental designs and will proceed with design-based inference techniques that aspire to approximate the experimental ideal without using randomized data. This course will comprise of theory lectures and lab exercises using R. By the end of it, students will be in a position to follow the relevant empirical literature and develop strong research designs testing their own causal hypotheses.
Fieldwork is a critical element of qualitative research across the social sciences. In this course, participants will be equipped with approaches, techniques, and practical tools to carry out fieldwork in Political Sciences and International Relations, with a particular focus on complicated field sites, such as conflict and otherwise fragile settings. The course will cover three main areas. First, we will explore the utility of ethnographic methods and approaches in Political Sciences and International Relations as well as the role of reflexivity as an integral part of the fieldwork process. Second, participants will learn about a variety of data collection methods, particularly semi-structured interviews and participant observation. We will also discuss techniques to address challenges that arise in the context of collecting data, such as source triangulation, dealing with missing data, establishing trust relationships, and mitigating access problems. Finally, the course will address matters related to risk management, safety protocols, and the researcher’s self-care during and post-fieldwork.
How will leaving the European Union affect the British economy? Who will win the next French presidential election? How long will the civil war in Syria last? How many people will migrate next year? This course introduces participants to ways to build forecasting models, to forecast future outcomes, and to describe the performance of forecasting models. Such models allow researchers to conduct credible test of theories and to satisfy their and the public's natural curiosity of what will happen next. The goal of the course is to familiarise participants with the logic of forecasting with statistical models and with the practical steps for building and evaluating such models.
Qualitative researchers put less emphasis on patterns of co-variation between variables, and prioritize the analysis of processes leading to specific outcomes. In fact, because of the limited number of cases that qualitative researchers study, the strongest empirical foundation for making valid causal inferences comes from the analysis of causal processes within cases. The N is usually too small to make more general statements about relationships among variables within a population. The analysis of causal processes within cases allows us to establish temporality and the direction of the causal relation under scrutiny, as well as to test theories implying causal chains and mechanisms. This technique is often referred to as “process tracing.” Process tracing is more than simply describing events in the order they happened. In this short course, participants will learn about the central aspects of this method. First, the course will begin with a discussion of the nature of causal mechanisms and the philosophical foundations of a mechanismic view of social science. We will then explore the contrast between explanatory, stepwise theories of mechanisms, which are a requirement for rigorous process tracing, and those that focus on predictions about constant conjunctions. Second, the course will discuss how to derive observable implications from theories that imply causal chains and mechanisms, and how to probe these implications empirically. To this end, the lectures will offer an introduction to the design of different types of process tracing tests. Crucially, participants will learn to evaluate the strength of these empirical tests using the Bayesian framework. By the end of the course participants will have the conceptual and methodological tools to craft research proposals that serve as roadmaps for data collection and analysis in the field.
Why let them read your data if they can see them? Visualising data is a powerful strategy for creating reports and conveying clear messages to readers, but traditional social sciences training does not cover data visualisation. This course will introduce participants to the use of graphs to report and show description, variation and correlations of data, and we will use graphical representation of effects, simulations and quantities of interests in order to report statistical results in a more straightforward way. As well as covering the basics of how to draw an effective figure or graph, we will also work on drawing maps and combining graphs. At the end of the course participants will have a wide choice of graphs for use in their future research. Participants are asked to submit a visualisation project with its related data before the course. This can then be discussed with the instructor, and developed as part of the course. Visualisations will be done in R, and when possible, Stata codes will also be provided.
Additional Sessions (Optional)
A daily walk-in clinic on ‘R’ will be available for an hour during the lunchbreak.