data analysis – Cynthia D'Angelo

Toward a more open science practice with R

Recently I did a webinar with my colleague Joshua Rosenberg, hosted by the Center for Open Science, on Analyzing Educational Data with Open Science Best Practices, R, and OSF. You can find a recording of the webinar here and our slides and an example R Notebook are in an OSF repository here. I thought I would do this blog post to summarize some of the main things I talked about there and highlight some of the more important aspects.

This webinar was ostensibly about open science for educational data. I think most of us want to engage in more open science practices (which could include open data, open materials, preregistration of studies, replication, posting preprints, and reporting null results) but don’t know necessarily where to begin or what tools to use. I think we tried to make the argument that workflows, procedures, practices, and behaviors that are good practice for you by yourself, future you, and your internal team can also be good for open science. And that using R and its many packages and tools is a good way of achieving those goals.

I’ve written many times before about how much I love using R and how I want others to incorporate it more into their practice. I’ve now collected the series of blog posts as well as other related blog posts (like this one!) and slides onto one page for easy access. You can go to cynthiadangelo.com/r/ to see all of the R related stuff that I have worked on linked in one place.

In general, I’ve been thinking a lot lately about my values and commitments as a researcher and how I approach my work in a very basic way. What is important to me, my collaborators, and my field? How could I be doing things differently or looking at my assumptions differently?

This thinking led me to this set of things to consider for a more open science approach:

There are a lot of technical tools and solutions to some of the open science problems. But there are also philosophical/ethical/moral issues to consider.
Humans are participants that helped produce your data. All humans deserve respect and so do their data.
There’s no easy answer for some of these situations you might face. That’s ok. Part of what open science asks is to consider your options and document your decision making.
Reflect early on in your process about what your goals are and how you want to achieve them. What are your values? How do these match up?

Some of the tips and guidelines that I talk about in the webinar: (1) Documentation is so important. It’s also really difficult. Making things clearer for you and your future self will also make them clearer for others who might eventually see your code. (2) A tidy data structure will make things easier for you and easier for others to understand. If you’re not already on board the tidyverse train, it’s never too late to start. (3) Make sure you have a data dictionary somewhere that explains all of your variables. This sounds obvious, but it doesn’t usually happen because in the moment you think you understand all of your variables. But future you will not remember all of those things. Write it down. Preferably in a R Notebook (more on that later). (4) Pick a consistent and clear file naming convention early on in your project (preferably before you begin data collection). Think about the date format you use and think about the unit of analysis you will care about later and try to incorporate it directly into your filename to help with filtering and analysis later on. (5) Of course I want you to visualize your data. Descriptive statistics can be misleading sometimes and visualization is an important step in your process and is not just an end product.

The thing that ties all of this together is using a R Notebook within RStudio. R Notebooks make use of RMarkdown, a flavor of Markdown, my favorite way to write. It is a plain text file, so it’s easy to version control and easy to share, both things that are hugely important when thinking about open science. I really like R Notebooks because you can easily incorporate explanatory text alongside your code and figures/graphs are persistent across the page so you can scroll and easily refer back to something above or below where you are working. This, in my opinion, is a much better way to use R than the older way with scripts and the console.

R Notebooks can produce an html file that you can send to your colleague or friend who doesn’t have R installed and they will be able to open it up in a browser and see all of your wonderful thoughts and figures. It’s really great. You can also execute code in Python or JavaScript or D3 (or a few other programming languages) in addition to R, so it’s very versatile. There are a lot of output formats as well, including pdf, Word, slide decks, dashboards, and books. And they are all customizable. Check out the RMarkdown website to see all of the options and more details on how they work. For me, they dramatically changed (in a good way) how I do my work.

Maybe a good question to leave you with is to try and answer “What is the best way for you to work toward open science?” It doesn’t have to be a big thing; it can be a bunch of small changes over time. This hopefully shouldn’t feel too overwhelming. There are lots of us here to help.

Data Visualization

This is the fourth part in an ongoing series on how and why you should be using R, especially if you are a social science researcher or education researcher, like me. If you missed the earlier ones, you can check out part 1 (Intro to R), part 2 (R Basics), and part 3 (Data Cleaning and Manipulation). This post will go into some more specifics relating to data visualization.

There are many ways to visualize your data using R. By far the most popular (and I think robust and flexible) is using the ggplot2 package. This post will talk a bit about why and how to visualize your data and some tips and basics to using R’s ggplot2 package to help you achieve your visualization goals.

Why visualize?

There are lots of reasons why you might want to visualize your data (or rather, why you should visualize your data). It can be a useful tool at various stages of research, and depending on where you are in your analysis process, different aspects of visualization might be more or less important to focus on. The way I see it, there are three main purposes for data visualization: examining your data, showing your data/findings, and sharing your data/findings.

What question are you trying to answer with your data? How can a visualization help you answer that? Do you have a really complex data set that is too hard to easily capture with a few numbers? Are you interested in variation and distribution rather than just means and medians? Are you exploring different relationships between variables and want to see how they interact?

Data Cleaning and Manipulation/Organization

This is the third part in an ongoing series on how and why you should be using R. If you missed the earlier ones, you can check out part 1 (Intro to R) and part 2 (R Basics). This post will go into some more specifics relating to data cleaning, organization, and manipulation.

In my opinion, the dplyr package is a game changer for those trying to learn R. It is what motivated me from just recommending that people use R to basically demanding that my friends and co-workers switch to R. I remember the day that I finally got around to learning how to use the package’s functionality and all of the ways in which it lets you easily and clearly manipulate your data frames¹. I just kind of stared at my computer screen and imagined how much better my data-life was going to be with these tools. I realized that the hours and hours I used to spend in Excel trying to massage my data into the right form were over². Also, I wouldn’t have to decipher weird R base code anymore when trying to create new variables or filter datasets. The dplyr package and its friends make your code/scripts much easier to read which will help both you and future you in trying to decipher what is going on.

Continue reading “Data Cleaning and Manipulation/Organization”

10,000 Tweets

I have been on twitter for almost ten years. Twitter has changed a lot in that time and my enthusiasm for the platform has waned a bit over the years, but I still find it to be a compelling communication platform. Initially I used it to share about the more mundane, personal parts of life and my stresses as I finished graduate school. Lately it’s become more professionally-focused (most of the time) and more reflective of the many things that are happening in the world (but with important dog pictures also). I have met lots of people through twitter as well as listened and learned from thousands of people who I would never have met in my day-to-day life. It has helped me gain a wider audience for my academic work and has allowed me to share pictures of my awesome dog with strangers and friends alike.

I just hit 10,000 tweets (if I did this correctly then the tweet linking to this post would be number 10,000). And I thought it would be a good opportunity for me to go back through my twitter archive and get a sense of what all of those tweets were about and how I tweeted. (The analysis that follows is actually only on my first 9,945 tweets because I had to request my tweets a couple weeks ago and do the actual analysis.) This was also a fun R exercise for me¹.

Continue reading “10,000 Tweets”