Introduction

R is a growing, vibrant programming language. Originally created by Ihaka and Gentleman in the 1990s, what was a niche experimental language is now in mainstream usage across many disciplines and domains (Ihaka, 1998).

One thing that Ihaka and Gentleman may not have intended at the time was the fact that R is no longer just a programming language. Users of R are developing into a distinct programming subculture at the intersection of multiple domains. This places it in a unique position, both to benefit from and to influence those domains. For instance, ecologists interact regularly with economists in this community. While a software developer in Denmark can extend the concepts of a New Zealand statistician and plot seals swimming in the Antarctic ocean.

There are few communities where a wealth of intellectual diversity this large have commmon mediums of communication, common shared values and identities. We bring to this community our experiences and knowledge. We take our learnings and explorations back to our own domains.

This piece argues that not only is R a developing subculture: that being a subculture it can transmit and develop ideas that have a fundamental economic value beyond the coded output it enables. This value generation is a powerful force that helps members of the subculture perform more efficiently and more creatively.

This piece also argues that not only does R as a subculture engender value, it creates, develops and demands values. A subculture developing in R does not just develop and transmit ideas, it develops and transmits an ideaology, one that has the potential to reach well beyond the fields and domains of its userbase.

Not just a language

R is a programming language with two decades of development. In addition to that programming development, community development has expanded rapidly in recent years. The R community is not bounded by any single institution, group or online platform. While there are large and distinct user groups - online and on the ground: none of these wholly encompass the R community.

These communities, diverse and developing, share commonalities and crossovers that go beyond the programming language itself. They share distinctive patterns of behaviour, beliefs and mores.

These should not be construed as a wholly homogeneous set of beliefs and behaviours: indeed, there are many differences. R-Ladies has a distinctive difference to the R community on Stack Overflow, for example: even though they share some of the same purpose to promote R usage. These are both different to the #rstats Twitter community and distinctly different to meetups in cities across the world.

Anheier et al. (2009) notes that subcultures can exist at all levels of organisations. In other words, xkcd was right, it’s fractal all the way down (xkcd, 2018). However, these diverse groups share distinctive traits that tie them together.

If we take the definition of a subculture as per Hodkinson (2012), these distinctive traits are “…extra-ordinary in their location within the parameters and environment of unusually concentrated … communities centred on sets of styles and practices which, in some respects, are spectacular and distinctive and are understood as such both by insiders and outsiders.”

In the context of R, these are distinctive traits applied by others collectively (“the R community is …”) as well as those proudly championed within the community. For example, R’s own language quirks, notably the assignment operator, <-.

Haenfler (2014) defines subculture more specifically: “A relatively diffuse social network having shared identity, distinctive meanings around certain ideas, practice and objects, and a sense of marginalization of resistance to the perceived conventional society..”

In most respects, the R community fits very tightly into this definition.

Diffuse networks: There is no single organisation that controls the R community, though there are some that have a great degree of influence. There are, however, a number of institutions that R users regularly share with, participate in or are benefited by.

These include institutions critical to the functioning of R as a programming language, such as the R Core group, CRAN (Centralised R Archive Network) and the R Foundation. However, participants in the developing subculture are not limited to these institutions: many R users participate in other institutions like R-Ladies, or the R for Data Science learning community. Many user and interest groups are benefited by the work of the R Consortium.

Not every R user considers themselves a part of ‘the community’. In fact, many use R regularly without giving thought to a broader user community at all.

Shared identity: What makes a person a part of this developing subculture? I would posit it’s seeing a difference between R as something you use and R as something you share, in whatever format that is. There’s nothing wrong with seeing R in its intended purpose alone- just using is more than fine! But many people step beyond that and start sharing.

This is not limited to sharing code by writing packages or gists on github. It’s also as simple as sharing advice with new users, or being that new comer asking for advice. It’s taking the time to help a colleague write their first function, or pointing someone to a resource.

Shared distinct meanings: Many members of the developing subculture grapple with, embrace or dispute with certain philosophies, most notably ‘the tidyverse’. Indeed, there is a substantial differential of opinon within the subculture around the tidyverse, which paralells the experience Coleman (2012) observed among ‘hackers’ in the Debian project. Whether or not a member of the community is an adherent or a detractor: they maintain a shared meaning of what the tidyverse is in order to form an opinion on it.

The tidyverse should not be considered just a collection of packages. It is in fact a substantial subcultural phenomenon with a manifesto (Wickham, 2018), adherents and detractors. It has a manifesto composed of four principles. It is the last principle, design for humans, that underpins its purpose to improve accessibility of R usage to users without an extensive programming background.

Domain-specific issues, such as the replicability crisis in psychology (see for example Pashler and Wagenmakers, 2012) have created deep ideological discussions and responses from within the R community that transcend the original domain. Indeed, one of the most important institutions in the subculture, ROpensci, is predicated on improving the tools available to the subculture to manage reproducibility generally ( ROpensci, 2018).

Intellectual constructs like the layered grammar of graphics (Wickham, 2010) or the grammar of data manipulation (Wickham et al., 2017) contribute to overall meta constructs around programming and data agency.

Resistance to a mainstream culture: The R community was never formed as a part of a resistance to anything in particular. Ihaka and Gentleman were looking to build something better than what was available at the time. That nuceleus of code formed a community around itself over decades. As a subset of a broader open source subculture, the resistance to corporate control of source code is well known, especially as discussed by Coleman (2012).

Marginalisation: The final part of Haenfler’s definition refers to marginalisation. Many subcultures form around marginalised members of our broader society: queer subcultures, subcultures supporting people of colour to name two obvious examples. R users are not necessarily marginalised by using R, though many of them are due to other parts of their lives. R users are different to a broader, less keenly numerate and technically literate population. This may have led to some degree of marginalisation in the past. The proportion of R users picked last for bin ball in the playground may be distressingly high! However, at this point in time, in this society, the power of individuals who have agency over data and are literate in its meanings is increasing rapidly.

The developing R subculture has a number of facets of a shared lifeworld that ties these diverse groups together. These include:

Content developed especially for consumption within subculture. The developing subculture produces more than code for the use of analysts and analysis produced by R code. The developing subculture produces, both for its own consumption and those outside of it:

Visualisation as art: Art created with R alone or in combination with other tools is well-known (see for example the R Graph Gallery, Holtz (2017) or Flowing Data (Yau, 2018)).

But it can also be produced accidentally and shared for appreciative purposes (see for example accidental aRT (Woo and Mudrak 2018)).

Analysis for entertainment and general education: Many analyses produced in R by the subculture are not intended solely for serious scientific or business purposes. They can be intended to educate about tools or techniques or broader social issues. Or they can be simply created for the pleasure the maker has in the act of analysis and sharing. Two well-known examples include Robinson’s (2016, 2017) analysis of Trump’s tweets and Meek and Averick (2017) (although the site was built in D3, analysis was also conducted in R).

Extensive educative materials and communities: The purposes for these materials are varied, but include materials for R, the programming language (see for example Bocek (2018), Maegan (2018) among many others), for the dynamic package sharing system, such as Silge (2017) and for navigating the community itself, such as Ellis (2017) as well as many other specific purposes.

Meta cultural content: Created for the purpose of alerting the subculture to resources, developments or other community concerns, examples include R-Bloggers, R Weekly, Rseek and the R Journal.

This meta-cultural content need not be formalised. One prominent informal example is Mara Averick’s use of the Twitter platform to inform and engage with R users who may not necessarily be a part of a tight-knit R community.

Averick makes use of informal methods of communication, such as the use of emoji. These signal to readers that the information to be conveyed is accessible. Communication of highly technical information can be difficult if users do not have formal training. This technique has been highly successful in spreading content beyond the core subculture (Averick, 2018).

A well-developed lifeworld is not unique to R as a developing subculture among programming languages. The seminal work of Coleman (2012) first noted the development of a rich open source subculture within the Debian project.

We have established an R subculture is developing. Does a distinct developing R subculture matter? If so, what does it mean?

Subcultures have the power to create value

A programming subculture is more than just syntax: it’s a confluence of institutions, inputs and a repeatable cycle of output and value generation. The development of a subculture is driving the development of R, the programming language and in reverse the same is true.

A subculture is a powerful means of generating and transmitting information. Subcultures have both formal and informal institutions that generate value and disperse it both within and beyond the subculture. A common R subculture between disciplines helps reduce the intellectual isolation many work under and promotes a marketplace of ideas where good ones get better and great ones move far beyond their initial disciplinary home.

The fact that open source has value is not new: consider for example, the Debian project and Microsoft’s original opposition to FOSS (Free Open Source Software) and their modern embrace of it, most recently demonstrated by the acquisition of Github (Warren, 2018). But it is not just the technology being created within the R subculture that is valuable.

In a subculture, there is an evolutionary tourney of ideas. A subculture absorbs ideas from the outside, sets them to compete with each other, evolves them and generates new ones. Being at the intersection of multiple academic disciplines (statistics, ecology, biology, economics, engineering, computer science, sociology, politics to name a few), schools of thought (e.g. sciences, business, humanities) as well as practioners (academics, students, business, government) means an incredible confluence of ideas absorbed and interacted with.

A common subculture gives ecologists and economists both a medium and a platform for communication that would not otherwise exist. A world-wide subculture spanning multiple spoken languages gives rise to interactions between users on different continents, both within the same fields and beyond it. Each of these participants brings with them their own perspectives, methodologies and use-cases. These combine to create a whole greater than the sum of its parts.

The evolutionary tourney of ideas is sometimes explicit and conscious: see for example, the development of the layered grammar of graphics. While the original R implementation had a single creator, Wickham (2009), it is now contributed to and built upon by hundreds. The Github repository associated with it has 152 contributors at the time of writing. While additions such as plotly (Sievert et al. 2017), Pederson (2018) and Wilke (2018) add up to many more.

At other times the tourney of ideas is less formalised. The importance of high quality documentation, for example, was an idea that has developed broad acceptance over a long period of time.

The participants in this tourney are not limited to scientific researchers and software developers: users, students of R and people with interest rather than work in the area can all contribute. The mediums are many and varied: talks given at local meet ups, to international conferences, Twitter and Slack conversations. All contribute to build shared concepts and ideas, as well as to challenge those currently accepted.

The value being driven by this tourney of ideas could be considered as of three different types:

Methodological sharing. New methodologies (as well as old ones not well known outside their specific fields) are picked up, transmitted and pollinated across new fields with incredible rapidity within this subculture. The sharing of new methodologies both within and without academia increases the value of scientific contributions from academia, as well as the value of outputs created beyond these institutions. By making more methodologies available to more practitioners for testing and exploration, the community can create more valuable end-products with better accuracy and usability.

Ideological sharing. Ideologies within the developing subculture are floated, debated, evolve and take root. These can include ideologies around diversity, ethics and robust science as well as methodologically specific ones, such a Bayesian concepts.

Strategy sharing. Shared and devloping concepts around workflows, professionalising software development management techniques, programming norms and statistical/methodological skill contribute to a more robust workforce with a broader outlook, doing better analyses, faster and more productively.

Subcultures have the potential to create values

The existence of a developing subculture has the power to create value: both economically and in terms of art, humour and community. However, the confluence of ideas promoted by a developing subculture also brings with it values. A subculture transmits not just its ideas but its ideologies. As a subculture, we have a shared common set of values. The subculture is constantly initiating new users into those values as they start to learn and use R.

Coleman (2012) notes that it is not a monolithic idaology in the hacker subculture exemplified by Debian. Nor is it in R. To quote Coleman, “Once we recognise the intimate connection between hacker ethics and liberal commitments and the diversity of ethical positions, it is clear that hackers provide less of a unitary and distinguishable ethical position, and more of a mosaic and interconnected, but at times divergent ethical principles.”

R as a subculture is dedicated to two broad ideas in particular: data agency, the ability to manipulate data at will, and data literacy, the ability to interpret the information it represents by various means.

We are at a unique place in history as a subculture with a viewpoint focussed on data. The data sciences are a powerful influence on a changing world. The potential of a subculture is that multiple diverging view points about a single development are possible. The potentiality of a subculture is to provide a robust, nuanced and debated view of new developments.

As a subculture, we are aware of both the value of the new, as well as the need for values to utilise the new fairly. This is particularly critical in light of the rapid development of AI technologies in our culture at large. For example, the Google Voice Assistant’s demonstration raised both admiration for the technical skill and productivity potential it displayed, as well as concerns about the ethics of a human-sounding bot (The Straits Times, 2018). Both are valid reactions to the technological development: there is value in the technical creation, but our values around our interaction with it are still developing. A highly technically literate subculture has the potential to grapple with both of these points, helping to inform a greater super culture.

Other ethical concerns of great import for the R subculture and the super culture beyond it include the rise of algorithmic decision making and its possible consequences. See for example McQuillan (2018) and Kehl, Guo and Kessler (2017).

The phenomenon of a subculture influencing a broader general culture in which it exists was well documented among the Debian hackers by Coleman. She noted, “By turning Linux and open source into household names, many more people learned about not just open source but also the ethical foundations—sharing, freedom, and collaboration—of free software production.” The Debian subculture directly participated in transitioning the view of open source software from unvaluable, unethical, unimportant, to the vibrant, essential and ethically informed technology it’s considered today.

R as a subculture has the potential to do something similar in the future. We are at a crossroads where algorithmic decision making, data-driven automation and AI are all a part of our near futures. A strong subculture debates new advances, sees the economic value in them and develops an ethical framework to demand that that this value be distributed fairly, while ensuring the costs are not borne unfairly by a minority.

Ethical explorations are active in the R community. These include projects developed with the assistance of the ROpenSci foundation such as that owing to Bailey (2018), and those created by individuals under their own ethical imperatives, such as those owing to Keyes ( 2016 and 2017). Keyes (2017b) also explores how community management can create (or fail to create) an environment of diversity and inclusion: this has direct ethical implications for a community at large.

The R community does not develop its ethics in isolation from the broader programming or the data sciences communities. The community is also exploring confluences of consequences through online media such as Twitter. In one such example, Thompson and Nemeth (2018), neither of whom appear to be R programmers, explored unintended ethical consequences in a post widely shared among data science communities, shown below.

These highly visible ethical explorations reinforce existing ethical norms within the R subculture, help develop new ones and educate new members of the developing subculture about the expectations of the community.

R as a subculture creates value, but it also receives, develops and ultimately transmits its values. There are some values that have taken root in this developing subculture and are points of pride for many within the community:

Open source: As an open source programming language, the ethics of open source are ‘built in’ to the developing subculture. Many see contribution to open source development of R as part of their identity within the subculture. They believe in creating new utilities for others to use and build on freely. Many of the critical institutions of the developing subculture, such as the R Foundation, CRAN and the R Consortium champion this value stridently.

Diversity and inclusion: There is a developing awareness in the R community that not all members of the subculture have equal access to a meritocracy. Many programs and organisations have stepped up to begin the work of changing this, notably R-Ladies and R-Forwards, amongst others. Many in the R community see their role in the community to be encouraging and welcoming to new users. Some go further and create diverse resources or host community events (online or otherwise) to support this aim.

Reproducibility. Originally a question of ensuring scientific contributions are accountable and testable, the importance of reproducibility now has much wider acceptance outside science and in the analytics community at large. Suppported by ROpenSci’s purpose of creating tools for reproducible science, the concept has also been championed by Leek (2017, 2018) and Leek and Peng (2015), amongst others.

The potential of the R community to influence a broader superculture has a precedent in the Debian subculture.

The R community is a group of individuals collectively identifying as part of a whole. Each one contributes his, her or their values to the collective. This is the case whether the decision is made consciously and followed up by action or if no conscious thought is given to values individually or collectively.

When conscious thought around values is missing, it is not the presence of an absence of values: rather an absence of the presence of thought for them. We should choose our values and champion them, or as Thompson and Nemeth (2018) observed, they may be chosen for us. These values need not and should not be monolithic: the tourney of ideas within the subculture supports ethics as a dynamic concept, changing in response to a changing world.

Members of the subculture take their values with them into their workplaces and use them to make judgement decisions about the work they are doing. They are in turn influenced by the ethics in their workplaces and bring these back to the R community. This has the potential to create powerful change in the culture at large.

Beyond syntax

The R programming language has a vibrant, developing subculture attached to it. This subculture creates not just programming utilities, but art, educational artefacts, communities and research across many domains and fields. The subculture has championed ideas that have had a profound and valuable impact on the development of a modern analytics practice, namely visualisation, workflow, documentation and interoperability.

The presence of a subculture allows ideas to be inherited from other subcultures, altered, debated and built upon in a kind of evolutionary tourney. This goes beyond content created and into ideology and values. As a community focussed on data agency and literacy, the developing subculture has the potential to be highly influential in the super culture in which it exists, being able to provide a nuanced, robust and multifaceted view of new technological developments.

Acknowledgements

This piece benefited enormously from input, suggestions and challenges posed by Jesse Mostipak, Di Cook, Kim Fitter, John Ormerod and Charles Gray. Thank you all.

References

— (2018): “A database of suspected London gangsters has been widely shared”, The Economist, 18/05/10, https://www.economist.com/news/britain/21742151-over-third-those-named-it-have-no-serious-convictions-database-suspected-london?frsc=dg%7Ce, accessed 18/05/23.

— (2018): “Human-sounding Google Assistant sparks ethics questions”, The Straits Times, 18/05/10, https://www.straitstimes.com/world/united-states/human-sounding-google-assistant-sparks-ethics-questions, accessed 18/05/31.

Allaire, J.J., Tang, Y., Geelnard, M., 2018: Reticulate: Interface to ‘Python’, R package version 1.5, https://CRAN.R-project.org/package=reticulate.

Anheier, Helmut K., Stefan Toepler and Regina List, eds., (2010): International Encyclopedia of Civil Society, Springer New York.

Averick, M. (2018): Twitter Post, https://twitter.com/dataandme/status/999834698997170178, accessed 18/05/27.

Averick, M. and Meeks, E., (2017): Beyond the Tip: A data-driven exploration of Archer, https://archervisualization.herokuapp.com, accessed 18/05/21.

Bailey, P. (2018): Ethical Machine Learning: A vignette designed to assist with spotting and preventing proxy bias, https://github.com/dynamicwebpaige/runconf18, accessed 18/05/23.

Blischak, J., Carbonetto, P., Stephens, M. (2018): workflowr: A Framework for Reproducible and Collaborative Data Science, R package version 1.0.1, https://CRAN.R-project.org/package=workflowr, accessed 18/05/31.

Bocek, M., (2018): Data Science Course in a Box, https://github.com/rstudio-education/datascience-box, accessed 18/05/21.

Brace, A., (2018): “GGPlot disaster/masterpiece/album cover!”, Accidental ARt, http://accidental-art.tumblr.com/image/166785763056, accessed 18/05/31.

Bryan, J., (2017): Stat 545, https://github.com/STAT545-UBC, acessed 18/05/22.

Bryan, J., (2018): Happy Git and GitHub for the useR, http://happygitwithr.com, accesed 18/05/22.

Bryan, J., Zhao, J., (2017): googlesheets: Manage Google Spreadsheets from R, R package version 0.2.2, https://CRAN.R-project.org/package=googlesheets.

Cairo, A., (2013): The Functional Art, New Riders, Berkley.

Cairo A., (2016): The Truthful Art, New Riders, Berkley.

Coleman, E.G., (2012): Coding Freedom: The Ethics and Aesthetics of Hacking, Princeton University Press, Princeton.

Cook, D., Lee, E-K., Majumder, M. (2016): “Data visualization and statistical graphics in big data analysis”," Annual Review of Statistics and Its Application, Volume 3, p. 133-159, https://www.annualreviews.org/doi/10.1146/annurev-statistics-041715-033420, accessed 18/05/22.

Eddelbuettel, D., Francois, R., (2011). “Rcpp: Seamless R and C++ Integration”, Journal of Statistical Software, Volume 40, Number 8, p. 1-18, http://www.jstatsoft.org/v40/i08/.

Ellis, S.E., (2017): “Hey! You there! You are welcome here”, ROpenSci, https://ropensci.org/blog/2017/06/23/community/, accessed 18/05/21.

Haenfler, R. (2014): Subcultures: The Basics, Oxon, Routledge.

Hodkinson, P., (2012): Beyond Spectacular Specifics in the Study of Youth (Sub)Cultures, Journal of Youth Studies, Volume 15, Issue 5.

Holtz, Y., (2018): The R Graph Gallery https://www.r-graph-gallery.com, accessed 18/05/21.

Ihaka, R., (1998): R: Past and Future History, https://www.stat.auckland.ac.nz/~ihaka/downloads/Interface98.pdf, accessed 18/05/21.

Kehl, D., Guo, P., Kessler, S. (2017): Algorithms in the Criminal Justice System: Assessing the Use of Risk Assessments in Sentencing, https://dash.harvard.edu/handle/1/33746041, accessed 18/05/31.

Keyes, O. (2016): “Ethics, scientific consent and OKCupid”, Ironholds, https://ironholds.org/scientific-consent/, accessed 18/05/23.

Keyes, O. (2017): “Stop Mapping Names to Gender”, Ironholds, https://ironholds.org/names-gender/, accessed 18/05/23.

Keys, O. (2017b): “Behaviour and Community in Open Science”, Ironholds, https://ironholds.org/not-about-image/, accessed 18/05/23.

Landau, W. (2018): drake: Data Frames in R for Make, R package version 5.1.2, https://CRAN.R-project.org/package=drake, accessed 18/05/31.

Leek, J., (2017): “Reproducibility and replicability is a glossy science now so watch out for the hype”, Simply Statistics, https://simplystatistics.org/2017/03/02/rr-glossy/, accessed 2018/05/23.

Leek, J., (2018): “The value of re-analysis”, Simply Statistics, https://simplystatistics.org/2012/12/18/the-value-of-re-analysis/, accessed 2018/05/23.

Leek, J., Peng, R.D., (2015): “Opinion: Reproducible research can still be wrong: Adopting a prevention approach”, Proceedings of the National Academy of Sciences of the United States of America, Volume 112, Number 6, p. 1645-1646, http://www.pnas.org/content/112/6/1645, accessed 18/05/23.

Leeper, T., (2018): “Release ‘open’ data from their PDF prisons using tabulizer”, ROpenSci, https://ropensci.org/blog/2017/04/18/tabulizer/, accessed 18/05/22.

Lin Pedersen, T., (2018): ggraph: An Implementation of Grammar of Graphics for Graphs and Networks, https://CRAN.R-project.org/package=ggraph, accdsed 18/05/30.

Maegan, J., (2018): R4DS: the next iteration, https://www.jessemaegan.com/post/r4ds-the-next-iteration/, acccessed 18/05/21.

McKinney, W., (2018): Announcing Ursa Labs: an innovation lab for open source data science, http://wesmckinney.com/blog/announcing-ursalabs/, accessed 18/05/22.

McQuillan, D., (2018): “People’s Councils for Ethical Machine Learning”, Social Media + Society, published May 2, 2018, http://journals.sagepub.com/doi/full/10.1177/2056305118768303#articleShareContainer, accessed 18/05/31.

Meager, R., (2018): Twitter post, https://twitter.com/economeager/status/997077103483981825, accessed 18/05/22.

Pashler, H., Wagenmakers, E-J., (2012): “Editors’ Introduction to the Special Section on Replicability in Psychological Science: A Crisis of Confidence?”, Perspectives on Psychological Science, Volume 7, Issue 6, http://journals.sagepub.com/doi/10.1177/1745691612465253, accessed 18/05/27.

R Core Team, (2017): R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna.

Robinson, D., (2016): “Text analysis of Trump’s tweets confirms he writes only the (angrier) Android half”, Variance Explained, http://varianceexplained.org/r/trump-tweets/, accessed 18/05/21.

Robinson, D., (2017): “Trump’s Android and iPhone tweets, one year later”, Variance Explained, http://varianceexplained.org/r/trump-followup/, accessed 18/05/21.

ROpensci (2018): “Our Vision and Mission”, ROpenSci, https://ropensci.org/about/, accessed 18/05/27.

Sievert, C., Parmer, C., Hocking, T., Chamberlain, S., Ram, K., Carvellec, M., Despouy P. (2017): plotly: Create Interactive Web Graphics via ‘plotly.js’, R package version 4.7.1, https://CRAN.R-project.org/package=plotly, accessed 18/06/10.

Silge, J., (2017): Navigating the R Package Universe, https://juliasilge.com/blog/navigating-packages/, acccessed 18/05/21.

Thompson, C., Nemeth, S. (2018): I Built Software to Kill People, https://twitter.com/stephaniecodes/status/992419444810010625, accessed 18/05/23.

Unwin, A., Hofmann, H., Cook, D., (2013): Let graphics tell the story - datasets in R, The R Journal, Volume 5, Issue 1, p. 117-129. https://journal.r-project.org/archive/2013/RJ-2013-012/RJ-2013-012.pdf, accessed 18/05/22.

Urbanek, S., (2016): rJava: Low-Level R to Java Interface, R package version 0.9-8, https://CRAN.R-project.org/package=rJava.

Warren, T. (2018): “Microsoft’s GitHub acquisition celebrated by the Linux Foundation”, The Verge 2018/06/08, https://www.theverge.com/2018/6/8/17441096/microsoft-github-acquisition-reaction-linux-foundation, accessed 18/06/10.

Wickham, H., (2009): ggplot2: Elegant Graphics for Data Analysis, Springer-Verlag New York.

Wickham, H., (2010): “A Layered Grammar of Graphics”, Journal of Computational and Graphical Statistics, Volume 19, Number 1, p. 3 - 28.

Wickham, H. (2018): “The Tidy Tools Manifesto”, Tidyverse, http://tidyverse.tidyverse.org/articles/manifesto.html, accessed 18/06/01.

Wickham, H., Bryan, J., (2017): readxl: Read Excel Files, R package version 1.0.0, https://CRAN.R-project.org/package=readxl.

Wickham, H., Francois, R., Henry, L., Müller, K. (2017): dplyr: A Grammar of Data Manipulation, R package version 0.7.4, https://CRAN.R-project.org/package=dplyr.

Wickham, H., Miller, E., (2018): haven: Import and Export ‘SPSS’, ‘Stata’ and ‘SAS’ Files, R package version 1.1.1, https://CRAN.R-project.org/package=haven, accessed 18/06/01.

Wilke, C., (2017): cowplot: Streamlined Plot Theme and Plot Annotations for ‘ggplot2’, R package version 0.9.2, https://CRAN.R-project.org/package=cowplot, accessed 18/05/30.

Woo, K., Mudrak, E., Accidental aRt, http://accidental-art.tumblr.com, accessed 18/05/21.

Yau, N., (2018): Flowing Data, http://flowingdata.com, accessed 18/05/21.

Yau, N., (2018b): “Career Paths”, Flowing Data, http://flowingdata.com/2017/11/28/career-paths/, accessed 18/05/31.