Spring School Introduction R

'An Introduction to R, including Data Management and Programming' and ‘A Workshop on Automating Cross-National Statistical Modelling’ David Armstrong and Raymond Duch (University of Oxford)

29-30 September 2008

The courses will be taught at the Department of Politics and International Relations, Manor Road Building, Oxford.

‘An Introduction to R, including Data Management and Programming’ will take place on Monday (29 September) from 9:30am-5pm. The 'Workshop on Automating Cross-National Statistical Modelling’ will take place on Tuesday (30 September) from 9:30am-5pm.

Please note also that registration for the workshop ‘Automating Cross-National Statistical Modelling’ is on a first-come, first-served basis and that the workshop is free of charge.

Registration for both days will close at noon on Monday 1 September 2008.

Tea, coffee and lunch will be provided on both days.

Please see below for a summary of each event.

Further information pdf

Programme pdf

'An Introduction to R, including Data Management and Programming', taught by Dave Armstrong

R, is an open-source (i.e. free) piece of statistical software that is the default statistical language in Statistics departments and is growing in popularity in the social sciences due to its flexibility and relative ease of use. The crash course assumes no prior knowledge of R, and will take participants from installation of the software to opening and editing data to running models. The course will also deal with the graphical display of data and model results. Additionally, the course will deal with different data structures and some basic data management tasks. Further, the course will offer an introduction to basic programming in R, including loops and other mechanisms for automating computations. The programming topics will be of particular interest to scholars attempting to pool data from a number of different sources as will be demonstrated in the workshop the following day.

This is an intensive one-day course, to take place on 29 September 2008

Dave Armstrong is Post-doctoral Fellow in Quantitative Methods, Nuffield College

‘A Workshop on Automating Cross-National Statistical Modelling’, led by Dave Armstrong and Raymond Duch

This workshop introduces methods for combining and analyzing a large number of different datasets with an eye toward testing hypotheses about individual- and survey-level political and social phenomena in different contexts. Traditional techniques for accomplishing this goal (e.g., multilevel modelling) require that all of the datasets have the same variables present. This often results in throwing a lot of information away. The set of techniques we discuss will attempt to use all of the relevant data from all of the different data sources to generate results. As more and more organizations are undertaking the task of generating election and other social surveys (e.g., WVS, CSES, BES, NES, LatinoBarometer, EuroBarometer, AfroBarometer), these techniques become increasingly useful. We have collected roughly 500 such datasets and have successfully implemented an automated strategy to simulate quantities of interest at the individual and survey levels over all of the datasets. We will provide step-by-step instruction on how to go about this type of work from theoretical issues, to data collection and preparation to generating and presenting the results. We use as a running example, the work done by Duch and Stevenson (2008) which uses these different datasets from different contexts to learn about the prevalence and magnitude of economic voting. In particular, these authors wanted to know how a change in economic perceptions might affect the likelihood of voting for the prime minister ’s party in the next election.

The workshop will talk about simulation as a method for obtaining quantities of interest with measures of uncertainty. By simulation, we are talking about a CLARIFY-type simulation (http://gking.harvard.edu/clarify/docs/clarify.html) where random draws are repeatedly taken from the coefficient vector and quantities of interest (like predicted probabilities) are calculated for each draw (King, Tomz and Wittenberg 2000; Tomz, Wittenberg and King 2003). The result is a distribution of the predicted probability that can be assessed for its difference from zero or from any other distribution.

Students Participants attending this workshop should be comfortable with linear and some basic generalized linear models (e.g., logit, probit). Further, students participants should have taken the course ‘Introduction to R, including Data Management and Programming’ offered the previous day or have familiarity with that material in R. In the interest of full disclosure, we are not offering the roughly 500 datasets compiled by the instructors. We will offer some data for practice, but these will be smaller versions of a few of the datasets used in these large projects.

This one-day workshop will take place on 30 September 2008.

Dave Armstrong is Post-doctoral Fellow in Quantitative Methods, Nuffield College

Raymond Duch is Reader in Quantitative Political Science, Nuffield College

Instructors:

Dave Armstrong is currently a post-doctoral fellow in the Department of Politics and International Relations at the University of Oxford. His substantive research focuses on the causes and consequences of state repression, particularly democratic institutions, though he is also working on models for manifesto-based party placements (with Jeremy Albright and Ryan Bakker) and strategic voting (with Ray Duch and Jeff May). He teaches quantitative methods at the University of Oxford, the Inter-university Consortium for Political and Social Research (ICPSR) Summer Program and the Essex Summer School in Social Science Data Analysis and Collection. His main methodological interest lies in measurement, though he is also quite interested in graphical presentation of results and non-parametric regression models. He has recently accepted an appointment as Assistant Professor of Political Science at the University of Wisconsin - Milwaukee to start in August of 2009.

Raymond Duch is the University of Oxford Professor in Quantitative Political Science and a Professorial Fellow at Nuffield College. Prior to assuming this position he was the Senator Don Henderson Scholar in Political Science at the University of Houston. He received his BA (Honours) from the University of Manitoba in Canada and his MA and PhD from the University of Rochester. In addition, he has held visiting appointments at the Hoover Institute and the Graduate School of Management, Stanford University, The Institute for Social Research Oslo and the Wissenschaftszentrum Berlin für Sozialforschung. Professor Duch is currently Associate Editor of the American Journal of Political Science; recently served on the 2004 Planning Committee for the American National Election Studies; and was recently a member of the National Science Foundation Political Science Advisory Panel.

His major teaching and research interests are in the areas of quantitative methods, comparative political economy, public opinion research, and democratization. His articles have appeared in leading international scientific journals.

Professor Duch’s contributions to quantitative methods include econometrics, multi-level modelling, experiments, applied game theory, and survey research design.

At Oxford, Professor Duch is on the executive committee of the Centre for Research Methods in the Social Sciences (ReMiSS) and on the executive committee of ESRC Oxford Spring School in Quantitative Methods for Social Research the Spring Methods School.

Costs:

The 'Workshop on Automating Cross-National Statistical Modelling' – is free. Registration is on a first- come, first-served basis.

'An Introduction to R, including Data Management and Programming' course-

• Participants from ECPR-member institutions (both UK and abroad):£75

• Participants from non-ECPR academic institutions: £100

• Participants from non-academic institutions: £150

For participants from ECPR-member institutions (both UK and abroad): these participants are also eligible to apply for travel and accommodation costs through the ECPR's mobility fund, and for fees through its scholarship fund. Details on the ECPR mobility fund.

For UK university and HEFCE-funded staff: in order to stimulate the uptake of high quality training courses the ESRC Research Resources Board makes available 50 bursaries each year for up to £ 1,000 each to enable HEFCE-funded staff in the UK engaged in teaching methods or supervising research to update their skills. Contract researchers working in HEIs are also eligible to apply. For further details and an application form, visit: http://www.ncrm.ac.uk/about/funding/training/

Registration:

Registration for this course and workshop is now closed.

If you have registered for the workshop and need to cancel your place, please notify Kate Candy as soon as possible so that the place can be made available to another participant.

[top]

Accommodation:

Trinity College: There are 35 single rooms available on a first come, first served basis for 3 nights (28-30 September inclusive). Accommodation is in study bedrooms usually occupied by students in term time. All rooms are equipped with a hand basin, telephone, refrigerator, hospitality tray with electric kettle and Ethernet socket for internet connection. There are toilets and showers on every staircase.

The cost of overnight stay at Trinity College, including full English Breakfast, is £47.70 per night. This is payable in advance of the course / workshop, when your place is confirmed.

Please indicate on the registration form if you would like accommodation during the course.

If you would like to make your own arrangements to stay in Oxford, you can find information about accommodation on the Oxford City website or on the Oxford Daily Information website.