In one of my previous articles, I talked about transitioning from psychology (or any social science) to data science. The focus was mostly on the skills one needed to gain to become a fully-fledged data scientist.
However, what if you already had made the transition? What could you do to leverage your existing psychological knowledge as a data-driven professional? It would be such a shame to throw-away years of studying and ignore all that you have learned!
I truly believe that psychologists have specific skills that can be used to become great data scientists! Sure, as a psychologist I am biased, but let’s just ignore that for a moment 😉*
This article will focus on identifying which psychological skills can be used, and how, in your work as a data-driven professional.
This will not be as self-evident as a tutorial. The purpose here is to inspire you to look for creative ways to be both a psychologist and a data scientist in your work. I hope to achieve this by providing you with some clear examples that might trigger you to apply old psychological methods in a new way.
*NOTE: The skills in this article are not exclusive to psychologists or those with a social background. The focus is as such since chances are that a psychologist is likely a better communicator than a software engineer.
1. Psychological Skills
Let us start at the beginning, why would I think that a psychologist could make a great data scientist? Surely, it is not because they are great coders or because they have extensive experience with git, right? No! It is mostly due to two broad skills they typically have acquired:
Communication skills 💬
Research skills 🔍
Now you might think that this does not seem like much, and by looking at these two skills I would be inclined to agree with you. It does not seem very impressive… However, mastering these skills requires significant effort and the impact of using them can be greater than you might think.
Both skills are umbrella-concepts and, as such, contain several sub-skills that each contributes in its own way to the fields of psychology and data science.
Below, I will briefly go through some of these skills to give you a feeling as to why communication and research skills are more important to data science then you might realize.
Explain it like I am 5 (ELI5) is often used to ask for a simple explanation to a complex topic. If you can explain it to a 5-year old, you truly understand the topic and shows that you can communicate the concepts across all knowledge levels.
As a psychologist, you are already trained to do so. It is vital that your patients understand complex methods, such as cognitive-behavioral therapy, as it would be difficult to apply treatment if they do not.
Stakeholder management 💬
It is important to realize that companies are social organizations through meetings, conference calls, lunches, etc. Being able to communicate at an advanced level is of great advantage.
Especially when it comes to stakeholder management, truly understanding their problems, and thereby the solutions they prefer requires that you have a feel of who they are as a professional. Why is this person asking for this feature? What is it that they are trying to achieve?
Ethical Decision Making 💬
A psychologist should know about ethics regardless of your specialization. In the context of data science, this involves patient confidentiality, privacy, decision making, etc. Understanding and having to work in complex ethical environments helps in preventing biases such as the selection bias, confirmation bias, or response bias.
Experimental knowledge 🔍
Now those with a social background have learned a bit more than just communication skills. In most programs, a large portion of the courses is geared towards the understanding of experimental setups involving humans.
Rigor in experimental design, nuanced interpretation of results, intellectual curiosity, are all skills befitting of a good psychologist.
Interestingly, most psychology programs are much more statistics-heavy than you would expect. In my program, one-fourth of all courses were some form of statistics. This ranged from statistical tests and validating questionaries, to A/B testing and
Those studying social sciences tend to have more in-depth knowledge about experimental design and statistics than most other programs. Seeing as a great deal of machine learning stems from statistics, this is an awesome skill to have.
Subject Matter 🔍
When analyzing human behavior, such as designing experiments for A/B testing or interpreting geographical data of individuals, then it would definitely be helpful to have a good grasp on the things that drives humans.
Psychologists can of great help when analyzing data pertaining to human behavior!
2. Type of Data Professionals
Although the skills above are important, they might be less or more important depending on the work you actually do. Thus far, I have been looking for the perspective of a data scientist. In practice, the work between data scientists might differ greatly as professions within the data-domain are still ill-defined.
In order to understand how psychology can contribute to a data-driven professional, it is important that you define what kind of professional you want to be!
In my experience there are roughly 7 types of data-driven professionals*:
Manager (managing a team of data-driven professionals)
Business Intelligence Consultant
Machine Learning Engineer
These roles significantly differ in the need for both technical and psychological skills. A manager, for example, typically would not need as much technical knowledge as an AI researcher but should definitely be more psychologically inclined in order to properly manage its team.
The data profession that suits you best can be condensed into a simple question:
What balance do you want between psychological- and technical skills in your work?
Answering this question might help you understand what kind of professional you want to be. If the answer to that question is that you are not looking for technical work, then most psychology professions would suit you (i.e., Clinical Psychology, Social Psychologist, I/O Psychologist, etc.).
However, the question becomes more difficult if there is some defined amount of both psychological- and technical skills you would like. The answer…a quadrant!
3. The Quadrant for Psychology in Data
The extent to which your psychological skills actually are helpful greatly depends on the kind of work you do. If you work as a data engineer and are mostly focused on creating data pipelines, then it is less helpful and necessary to have these skills.
Below, I have created a quadrant that indicates the need for psychological skills versus the need for technical skills in which several professions are placed. Although it is a simplification of reality (who actually wants to see a 4-dimensional graph 😅), hopefully, this helps you in deciding the profession that best suits you.
As seen above, the psychological skills typically correspond to communication- and research skills. In the quadrant above, there is a bit more emphasis on communication skills in order to separate it with technical skills.
The technical skills are those that include a wide range of skills within the data-domain, such as programming, machine learning, MLOps, data architecture, etc.
Interestingly, when filling in the quadrants I started noticing a trend in the relative placement of professions. It seems that a curvemight exist between technical and psychological skills:
A higher need for technical skills often leads to a lower need for psychological skills and vice versa.
Low Technical and High Psychological
I have placed professions that typically require more psychological skills than technical skills in the lower right quadrant: Business.
Those in the business quadrant are more likely to have either a consulting/advising role or a managerial role. With a greater need for communication and leadership skills, BI consultants and managers fill are a great fit!
NOTE: BI consultants can be very technical, especially when they are also in charge of the underlying data model. However, typically, it is more about understanding business than it is about creating complex algorithmic analyses. The same holds true for many of the other professions in the quadrant.
High Technical and Low Psychological
I have placed professions that typically require more technical skills than psychological skills in the upper left quadrant: Engineer.
Those in the engineer quadrant are people with a mostly technical background, like data engineers or machine learning engineers. Although it is helpful to have some psychological skills, it is not a necessity for the job as your interaction with stakeholders is less so than, for example, a data scientist.
High Technical and High Psychological
I have placed professions that typically require high psychological- and technical skills in the upper right quadrant: Unicorn(ish).
You do not often see people having both high technical and psychological skills as they are difficult skills to obtain independently, let alone together. You can see this in the quadrant’s population: there are no roles in the top-right corner.
Even the typical unicorn role of data scientists, in my opinion, does not need to have the same technical skills as an AI-researcher. They do need extensive technical and psychological skills, just not at the level of experts.
Low Technical and Low Psychological
I have placed professions that typically require little psychological- and technical skills in the upper right quadrant: Dataless.
Every quadrant needs a name, so I called this dataless for all roles that are not data related. I doubt there are any data-roles where you do not need both technical- and psychological skills.
NOTE: Since data-driven professions are typically ill-defined, the positions above should be taken as an indication and not as a ground-truth. Moreover, this is my perspective and is as such biases towards my experiences.
4. How to use Psychology in Data Science
Finally, the million-dollar question:
How can I actually use psychology in my work as a data scientist?
Although this can be done in many ways, I would like to focus on three aspects to get you started:
Analyzing human behavior
Communication with stakeholders is often very difficult due to the seemingly vague requirements they have for their projects. That is often not the fault of the stakeholder, but of the one asking for those requirements! It is up to you to gain an understanding of the stakeholder’s intentions when writing up those requirements.
It starts by going beyond asking for requirements. For each requirement, ask the stakeholder what he/she wants to accomplish with that requirement. Why do you want to predict next-year sales? What do you want to achieve by having that knowledge?
Realize that there is a lack of understanding on both sides. The stakeholder has little knowledge of what you do and vice versa. Crossing that bridge requires you to start identifying with the stakeholder.
NOTE: What helps me when analyzing communication errors between me and the other is assuming that I was in error. By assuming it was due to a fault of my own I can start identifying how I could have approached the situation differently. If you blame the other person for communicating poorly very little is going to chance next time around.
Great communication skills can require several years to develop but focussing a few aspects can help your skills tremendously. Let us focus on a single form of communication: presenting.
Presenting the results of your latest analyses can be more difficult than you might anticipate. You need to communicate your results, technical assumptions, impact on the business, validation metrics, etc. All of that while being interesting in the process.
Here are a few tips to get you started:
Know your audience
Keep It Simple, Stupid (KISS)
Explain it Like I’m 5 (ELI5)
Approach it like a story
Focus on a single message
Keeping it simple by focusing on a single message and explaining it in ways most can understand your results is key in communicating well.
Even for a technical audience, it is often best to start simple. Diving into very technical content from the start requires a tremendous amount of concentration. For technical individuals, you want to create intuition about what you are presenting and the methods you have applied.
Analyzing human behavior
Interpreting the results of your analyses often requires some degree of domain knowledge. As a psychologist, human behavior is your domain knowledge. However, there are not many organizations out there that focus solely on analyzing human behavior.
Fortunately, most organizations have some form of data pertaining to human behavior. For example, this could be customer reviews, tickets, or even hr-data. As all organizations nowadays have a website, you can, at the very least, analyze how people use an organization’s website and develop a behavioral flow.
You do not have to look very far to find behavioral data in any organization. Identifying these use-cases or problems can be beneficial as it allows you to show off your psychological domain knowledge!
NOTE TO END ALL NOTES: Although this article has gotten longer then I intended, it is still missing quite a lot of information and footnotes. I am making some assumptions here and there in order to compress information but it sometimes feels like that the un-compressed information would essential. I realize that I am generalizing at times, so please correct me if I am missing vital information or if I generalized too much. Finding a balance has proven to be quite difficult 😅.