Skip to content

docs: editorial pass on README, write-up, and About bio #1

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
savvides merged 3 commits into from
Apr 27, 2026
Merged

docs: editorial pass on README, write-up, and About bio #1

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
ba4b664
docs: editorial pass on README, HOW-IT-WAS-MADE, and About bio
savvides Apr 27, 2026
b00a979
docs: humanizer pass — drop performative reveals and rule-of-three
savvides Apr 27, 2026
1f67113
docs: drop inaccurate bio from About Philippos footer column
savvides Apr 27, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions HOW-IT-WAS-MADE.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,11 +10,11 @@ That's the fate of most dissertations. Years of work, compressed into a PDF, fil

## The Realization

In early 2026, I revisited my data with fresh eyes and modern tools. I found something the original analysis missed: an ecological fallacy hiding in the results. A correlation that looked real at the group level vanished within each condition. That was a genuine discovery, buried in an eight-year-old dataset.
In early 2026, I went back to the data with newer statistical tools. The original analysis had missed an ecological fallacy in the results. A correlation that looked real across the whole sample vanished within each condition. A real finding, eight years late.

I wanted to share it in a way that did the finding justice. Not as another paper. Not as a slide deck. As something people could experience.

Around the same time, I started using [Claude Code](https://claude.ai/code), Anthropic's AI coding tool. What I realized was that I didn't need to know how to build a website from scratch. I could describe what I wanted, and build it through conversation.
Around the same time, I started using [Claude Code](https://claude.ai/code), Anthropic's AI coding tool. I didn't need to know how to build a website from scratch. I could describe what I wanted and build it through conversation.

## The Process

Expand Down
4 changes: 2 additions & 2 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ An interactive data story exploring VR, learning, and a surprising statistical f

![The ecological fallacy reveal — dots colorized by condition show the overall correlation disappears within each group](docs/screenshot.png)

An interactive data story built from a 2018 PhD study (n=108) that tested whether VR headsets improve learning. The answer was surprising: the simulation matters, not the headset — VR and desktop produced identical outcomes (Cohen's d = 0.23) while both massively outperformed passive instruction (d > 2.4). A 2026 reanalysis uncovered an ecological fallacy hiding in the original results: a correlation that looked real at the group level vanished within each condition.
The 2018 PhD study (n=108) asked whether VR headsets improve learning. They don't. VR and desktop produced identical outcomes (Cohen's d = 0.23); both crushed passive instruction (d > 2.4). What mattered was the simulation, not the display. A 2026 reanalysis added something the original paper missed: the presence-learning correlation is an ecological fallacy — real at the group level, gone within each condition.

[**View the Interactive Story**](https://savvides.github.io/dissertation/)

Expand Down Expand Up @@ -118,7 +118,7 @@ The full addendum with all citations is in [`dissertation.md`](dissertation.md#u

## 2026 Retrospective

In 2026, I revisited this study with fresh eyes — red-teaming the methodology and reanalyzing the original data ([`data.csv`](data.csv)) using modern statistical tools. The core finding is rock-solid: interactive simulations produced massive learning gains over passive instruction (Cohen's d > 2.4). But the theoretical claims around *why* could be made much stronger. Key issues:
In 2026, I went back to the original data ([`data.csv`](data.csv)) and red-teamed the methodology with newer statistical tools. The headline finding holds: interactive simulations crushed passive instruction (Cohen's d > 2.4). The theoretical story about *why* doesn't:

- **Confounded conditions** — The low immersion group received fundamentally different content (video/text vs. interactive simulation), confounding immersion with interactivity
- **Uncontrolled time on task** — VR group spent ~10 more minutes than control (25.9 vs. 16.2 min)
Expand Down
3 changes: 1 addition & 2 deletions index.html
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -128,8 +128,7 @@ <h3>How This Was Made</h3>
</div>

<div class="about-column">
<h3>About Philippos</h3>
<p>Philippos Savvides is an educational researcher with a PhD in Educational Technology from Arizona State University. His work focuses on how learners interact with simulations, virtual environments, and emerging interfaces.</p>
<h3>Philippos Savvides</h3>
<a href="https://www.linkedin.com/in/savvides/">LinkedIn &rarr;</a>
</div>

Expand Down
Loading