Intermediate R: Capacities for Analysis and Visualisation

Important! The content below might describe more than we can cover in a week. Bear in mind that this can, to some extent, be changed according to your interests. When you have registered, I will contact you to ask about topics in which you are most and least interested and, I hope, we'll be able to shape part of the class according to your needs.

Learning R takes time, and often the next best step after learning the very basics of this programming language and its most important characteristics (covered in R Basics) is to get a good overview of what analyses it is possible to do in R, and how to do them.

The details of each topic and method are almost limitless – and in the end, this is part of what makes R so great – but for getting to know R at an intermediate level, these might be of secondary importance. You do not need to know the details – they will come later – you just need to know what is possible, what to look for and expect, and how to ask the right questions. These will make the steps that follow, the steps that you are most likely to do on your own while working on your analyses, much simpler and smoother. And this is the purpose of this course.

The course covers many topics, which in their substantive breadth could fill several week-long courses. But what we focus on in class is simply their technical implementation in R. For this course you do not need to know in any meaningful depth what, for example, multilevel models, panel data analysis or any of the other covered methods are, although it would be good to have at least a basic idea. We will simply look at their implementation in R – what packages and functions to use and what data structure is implied – with only a few very basic comments about the nature of the method and the type of data. This is not a course in statistical methods as such, but in how they can be used in R. We cover much technical ground and very little substance.

The course begins with the general issue of how to work most effectively with R. We look at some principles of how to write good, comprehensible and efficient R code and how to structure your files and folders, so that your work is smooth. Thereafter we look at two related topics – simulations and bootstrapping – which are applicable across specific methods as means to learn and understand them (simulations) or to get a grasp of the amount of uncertainty in your data and the results you can get from the latter (bootstrapping).

In the next two and a half days we will go over the implementations of some of the most common methods of statistical analysis in R. We will look at how to do factor analysis and structural equation models, how to work with panel data and multilevel data and many related analyses (for a more detailed list, see the day-to-day schedule). The structure of all these micro-topics will be the same:

• We spend the very minimum amount of time refreshing the basic logic of the method – what kinds of questions and what kind of data is it meant for?
• We look at the main packages in R that provide the functionality to implement this method.
• We look at how to use them – what functions they provide, what arguments the latter require and the structure of the output they give us.
• When the analysis is done, it is important to get an overview of the results. We look at how to get a good summary of your analysis and how to create the basic visualisations to evaluate your models.

More or less each model comes with its basic options for plotting, but sometimes you want more. Therefore, towards the end of the course, we take a longer look at how to use ggplot2, an R package which allows you to make almost any kind of data visualisation you desire.

ggplot2 is like a language within R, and to learn it fully would take more than a week. So we will only dip a toe in the water to see what’s possible with this package and how to make the basic kind of plots – barcharts, linecharts, scatterplots and the like – as well as how to customise them to your liking and save them in pdf or some other format of your choice.

Often, the end user of your data or your analyses needs to perform basic data operations, analyses and visualisations on their own. For this purpose, I will quickly introduce Shiny, an R package that allows you to make, with relatively little effort, interactive web pages where users can have a look at the underlying data on their own. There are many other ways in R to bring your results to the web, but Shiny is perhaps the most common, and a good place to start.

Creating visualisations is one way – and perhaps the best way – to get your results out of R, but sometimes you also need good-looking tables of model output. In that case, an R package called stargazer can save the day. Copy-pasting numbers from R to a table that you have manually created in your document is time (sometimes countless hours) that could be well spent doing something better. Stargazer lets you export your model results from R directly into LaTeX or html format, and the latter can easily be transferred to Word or any other software you might be using.

Sometimes it makes sense to integrate your code and its output, including data visualisations, with text and other media (like images and other online sources of information). R offers R markdown and R notebooks, which let you create html, pdf or word documents that include your R code, as well as any text you want to add, all within R.

When you have covered the topics of this class – when you know how to work in R more effectively, what kinds of analyses it is possible to do and the ways in which you can get them out of R quickly and effortlessly – you will be well on your way to becoming a proficient user of R.

I will assume you have solid basic knowledge of R on a level comparable to Akos Mate's R Basics course, for which this course is best taken as a continuation.

I also expect a very basic knowledge of the most common statistical methods used in the social sciences: what are they, what do they do, and when they should be used.