Hi ,
Why do academic research papers always seem to find something groundbreaking - and data scientists tend to make boring findings sound revolutionary?
As an academic turned data
scientist, I first felt the pressure to produce "interesting" statistical findings as far back as my honours year.
It was my first research project and the capstone to my Bachelors degree. I was determined to make a splash with my debut into the academic world. I imagined publishing my first research paper before starting my first real job.
9 months later, when my ultimate findings showed "no statistically significant evidence" to support my hypothesis, I was devastated.
No matter how much my supervisor tried to convince me otherwise, a null result just didn't seem to cut it.
Fortunately, I was eventually able to swallow my pride and accept my disappointing findings - which still led to me graduating with a First Class degree. However, a quick Google search is enough to find numerous examples of misleading statistics in academic research.
Years later, I discovered I wasn't alone in
this struggle.
As Nicholas Kelly, author of How to Interpret Data, recently shared with me, there's often pressure in academia to find "interesting" results - leading to statistical analysis that's... let's say... creatively interpreted.
The problem, though,
is that this erodes trust faster than you can say "p-hacking" and turns you into the researcher who cried "significance".
Here's the thing...
This same pressure exists in the corporate world. Data scientists feel the need
to find something impressive to justify their role, especially when executives keep asking "Why can't ChatGPT just do this?".
The temptation to make boring data sound revolutionary is real.
Nicholas's advice? Be honest when the data is boring. Trust is your most
valuable currency, and once it's gone, you won't get those cool, impactful projects in the future.
Nicholas joined me in the latest episode of Value Driven Data Science to tackle the critical challenge of data interpretation.
Our conversation
reveals:
- The four primary challenges that make data interpretation so difficult [02:24]
- Why ChatGPT and AI tools are changing the data interpretation landscape [06:23]
- The "Five Whys" technique that ensures you're asking the right questions instead of wasting time on problems everyone already understands [17:32]
- Why successful data projects don't end
with presenting insights and what to do next [20:01]
Ready to transform how you interpret data?
🎧 Listen now on Apple Podcasts or Spotify, or click the link below:
Episode 70: How to Interpret Data Like a Pro in the Age of AI
Talk again soon,
Dr Genevieve Hayes.
