Anyone* could learn data analysis. Probably.
I've of course been thinking about data analytics and classes since I'm supposed to stop procrastinating writing out my logs analysis course slides any day now. So last week's post by Benn (below) declaring that most graduate degrees in analytics being scams kept hold of my attention this week.
Benn Stancil
I think that most of us who've been in the industry a while have had various forms of misgivings with analytics masters programs, though I doubt many of us would be willing to outright call them scams. When asked to give my opinions of various programs, I've always wound up responding with something like "umm, I guess they're not terrible, but there's plenty of better alternatives". I've heard similar statements from others.
Meanwhile, many years ago while I was between jobs in the 2010s, I wound up having a conversation w/ a bootcamp company about analytics courses. While nothing came out of the conversation, the person I was speaking to had been asking about who I thought would be qualified to take an analytics course and my response was a set of very basic general skills like analytical thinking skills, understanding of numeracy, knowledge of business processes, etc.. The person I spoke to then blurted out "wait, you make it sound like anyone could become an analyst".
That offhand statement, probably said at least half in jest, continues to stick with me. Why, yes, I do tend to act like anyone who wants to can learn analytics enough to be effective with it. I've always maintained that the fundamental skills of doing data analysis are extremely basic. Most kids coming out of high school have probably been exposed to all the basic skills and all they're lacking is the direction to apply those skills to a relevant problem. While there are always many more advanced methods that take deliberate study to achieve and apply, my fundamental belief about analysis is that research questions come first, methods follow.
Questions before methods
I've always maintained that I'm a pretty decent tutor, but would make a terrible teacher – I'm capable of helping someone who wants to learn something figure out how to learn something, but I'm completely at a loss as to how to inspire someone who is at best indifferent to learning. My philosophy as to who can become an analyst – effectively anyone who wants to – is basically an extension of this belief. The fundamental tools for making sense of data is simple enough that anyone with motivation should be able to attain them while all the advanced stuff is specific to certain tasks.
So the question of who can be an analyst, to me, is about people finding motivation. What are the questions these people want to answer using research methods? That's going to dictate what they need to learn more than anything. Since there's more research methods out there than any single individual could ever learn, it's pretty much nonsense for someone to become some kind of "pure methods expert". It makes as much sense as someone who collects tools from all across the world, becomes extremely versed in how all the tools work, but then never actually apply the tools in practice.
The primary way I know that people acquire methods is because they are studying some kind of research question. Social scientists usually learn to wield questionnaires and structural equation models because their field has determined those methods are capable of helping shed light on what they're interested in. Economists, physicists, neuroscientists, political scientists, epidemiologists and all other fields developed their field-specific toolboxes as a result of the lines of inquiry they're engaged in. They frequently add, adjust, and even discard methods based on how they help answer interesting research questions. As befitting the name, tools are mere tools that can be thrown aside when better ones come along.
It's only in programs like data analysis bootcamps and masters programs that have a tendency to come at things backwards – they teach a bunch of common tools and methods with the expectation that students can apply them to industry. It's the equivalent of going to a bizarre school that teaches everyone how to use screwdrivers and soldering irons (because big companies are asking for those), but not hammers.
When teaching methods without a driving question to answer, you lose out on a lot of important meta-thinking involved in choosing a method. How can you debate the pros and cons of different methods with respect to the research question? How do you talk about making changes to protocols due to unique biases and quirks of the data at hand? That whole discussion is one of the more critical aspects about doing any data work or research and it's a serious problem if students aren't pushed to wrestle with the ideas. It's when you don't have this critical thinking portion do you get the "here to help (with algorithms!)" comic from xkcd. It's how you get the perennial "physicist/economist applies a method to another field, publishes a 'surprising' result, and the actual experts ask 'so, what?'".
Put all this together and it's why I think I'm able to help tutor groups of people into learning how to do certain types of analyses – anyone who wants to find an answer to a question is going to be receptive to learning ways to get closer to the answer. People who are incurious, or indifferent to questions aren't going to get very far.
In the end, I think teaching analysis is equally about both the method and whatever question brought the student into the room. There's a good number of methods like experiments, quasi-experiments, basic statistical inference, or even basic data collection that will be useful across a broad spectrum of questions. We can and should teach those, but while we're doing that, I feel it's super important to challenge students to think about the question they care about and whether the tools they're learning can apply. What tweaks are needed to make sure the assumptions of our models line up with reality? What other ways can we arrive at similar conclusions? Those conversations are what we actually do as analysts, and short of modeling that process out in the classroom, it's hard to impart.
But anyways, I really need to start putting pen to slides.
... After I test out a couple of missions in Monster Hunter Wilds...
Promise.
Standing offer: If you created something and would like me to review or share it w/ the data community — just email me by replying to the newsletter emails.
Guest posts: If you’re interested in writing something, a data-related post to either show off work, share an experience, or want help coming up with a topic, please contact me. You don’t need any special credentials or credibility to do so.
"Data People Writing Stuff" webring: Welcomes anyone with a personal site/blog/newsletter/book/etc that is relevant to the data community.
About this newsletter
I’m Randy Au, Quantitative UX researcher, former data analyst, and general-purpose data and tech nerd. Counting Stuff is a weekly newsletter about the less-than-sexy aspects of data science, UX research and tech. With some excursions into other fun topics.
All photos/drawings used are taken/created by Randy unless otherwise credited.
- randyau.com — Curated archive of evergreen posts. Under re-construction thanks to *waves at everything
Supporting the newsletter
All Tuesday posts to Counting Stuff are always free. The newsletter is self hosted. Support from subscribers is what makes everything possible. If you love the content, consider doing any of the following ways to support the newsletter:
- Consider a paid subscription – the self-hosted server/email infra is 100% funded via subscriptions
- Send a one time tip (feel free to change the amount)
- Share posts you like with other people!
- Join the Approaching Significance Discord — where data folk hang out and can talk a bit about data, and a bit about everything else. Randy moderates the discord. We keep a chill vibe.
- Get merch! If shirts and stickers are more your style — There’s a survivorship bias shirt!