Today marks day one of my trip to Washington, DC. It has been a busy couple of weeks, the kind where your calendar looks like a Tetris game and your brain starts naming folders like “final_final_v7_reallyfinal.”

I wanted to write a few posts before leaving, but the usual end-of-season chaos won. So, here’s the quick snapshot of what’s been happening as I head into a DC week focused on psychometrics, measurement, and the practical side of statistics.

• Wrapping up a few analysis pipelines (and cleaning up code so Future Me doesn’t suffer).
• Final edits on write-ups where “validity evidence” is not just a phrase, it’s the whole point.
• Lots of conversations about tests, fairness, and what we can responsibly claim from data.
• Packing, re-packing, then realizing I forgot something obvious.

If you’ve ever worked in measurement, you know the feeling: the work is detail-heavy, but the goal is simple. We want better decisions, and that starts with better measurement.

Why Washington, DC

DC is one of those places where policy, assessment, and real-world consequences sit at the same table. I’m heading there to spend time in the psychometrics world in a way that’s less “textbook examples” and more “actual decisions people are making.”

I also love the contrast: psychometrics is quiet work, mostly. It’s code, models, assumptions, diagnostics. DC is not quiet. It’s a fun mismatch, and it forces clarity.

What I'm Working On

My focus this week is staying practical and honest. Here are the themes I keep coming back to:

• Measurement first: if the construct isn’t measured well, the “big” structural model story falls apart.
• IRT and item diagnostics: item characteristic curves, information, local dependence, and what happens when items misbehave.
• Fairness checks: DIF, group comparisons, and the difference between “statistically detectable” and “meaningfully harmful.”
• Model checking: not just fit indices, but residuals, misfit patterns, and whether the model is telling the truth or just sounding confident.

There’s a version of this work that becomes a numbers contest. I’m not interested in that. I’m interested in defensible measurement, transparent reporting, and models that can survive a skeptical reader.

Looking Ahead

Over the next few days, I’m excited for the nerdy stuff: the conversations where someone asks one sharp question and suddenly your whole model needs a rethink. That’s not failure, that’s the process working.

I’ll also be posting small updates as I go, the kind that translate the “under the hood” psychometrics pieces into something readable.

If you want to follow along or connect, here are my links:

• GitHub: github.com/fatiihozkann
• LinkedIn: linkedin.com/in/fatih-ozkan-a5b602158
• Instagram: instagram.com/fatiih.ozkann

- Fatih