One elementary school teacher of mine once said:

"In a football (soccer) game, the goalkeeper can be a very good player. He can make some incredible saves during the match. If he concedes a howler, people will remember him by that.

The forward can do everything wrong during the match. If he manages to score one goal, this is sufficient to justify why he was playing."

A team without a forward will have a harder time trying to win, but it is possible. A team without a goalkeeper will certainly lose. If you think into it, it's possible to conclude that the goalkeeper is more important than the forward. But football isn't rational; it isn't about fairness or importance. It is about emotions: forwards get all the visibility and fame because they get to do the "fun" part of the job.

Why people subject themselves to be goalkeepers then? Well, not every one is the same. In this game, if you, for any reason, aren't good playing with your foot, and still want to play, there is only one position you can take.

I think about this all the time because feel that in our work environment, managers are forwards and devs/scientists/technicians are goalkeepers.

I work with statistics/data analysis and web development. I study these subjects for almost a decade and now I have 4 years of practical experience.

This information is on my LinkedIn profile and from time to time tech recruiters contact me wanting to have an interview. I always accept because I find it a great way to practice interviews and talking in English, as it isn't my native language.

A remark that I always make to my colleagues wanting to start doing data analysis related work is that it may seem similar to development, but it's not. When you develop, your code work or not. It may be ugly, it may be full of security problems, but you almost always have a clear indication if things are functioning. It's possible to more or less correlate experience using a programming language with knowing how to develop.

Data science is different. You have to know what you are doing because the code will run even if you are doing something totally wrong. You have to know how to interpret the results and judge if they make sense. For this the mathematics and theory behind is as important as the programming language you use.

Ok, so I go to my first interview for a data science position. Then I discover that I will be interview by... a psychologist. A particularly old one. Yeah. Great start.

She proceeds to go through the most boring checklist of questions I ever saw. The first one? "Do you know Python?". At this point I'm questioning myself why I agreed to be interviewed. A few minutes later, a super cringy one: "Can you tell me an example of your amazing analytics skills?". I then proceed to explain what I wrote in the last two paragraphs to her. At this point is clear that she has no idea of what data science is and the company probably googled what they should expect from a candidate.

20 minutes later and the interview is over. A few days later I receive an email saying that I was not selected to continue with the recruitment process because I don't have enough experience.

In summary: an old psychologist with no idea on how data science works says I don't have experience on the subject based on a checklist that they probably google. The interview lasted less than 30 minutes.

Two weeks later another company interviews me, I gave basically the same answers and they absolutely liked what they heard. Since that day I stopped trying to understand what is expected from you on interviews.